Dear Founders,

Team Ignite values the chance to learn about what you are building. Each year, we review thousands of companies across B2B SaaS, AI, fintech, deep tech, marketplaces, infrastructure, and network-dependent businesses. Some become obvious passes quickly. Some look unimpressive at first and later become exceptional. Some look incredible in the pitch and collapse under diligence. The only honest response to that reality is not “trust your gut harder.” It is to build a decision system that is ambitious about upside, skeptical about weak evidence, and humble enough to learn from both misses and wins.

This guide is the founder-facing version of how we think.

It is not our internal scoring model. We are not publishing our proprietary AI weights, formulas, cutoffs, or proprietary decision rules. But we do want founders (and our limited partners) to understand the logic behind our evaluation, because the fastest way to improve your odds is not to pitch louder. It is to bring clearer evidence against the questions that actually drive conviction.

A strong pitch does three things:

1. It shows why this could become a power-law company.
2. It proves the company is ready for institutional capital at this stage.
3. It names the risks clearly enough that we can believe you understand the business you are building.

Most founders optimize for persuasion. The best founders optimize for proof.

The mindset shift: from narrative to evidence

A pitch is not a performance. It is a compact argument.

Every great startup pitch makes a small number of claims:

• This team is unusually capable of winning this market.
• This problem is severe enough that customers will change behavior.
• This moment creates an opening that did not exist before.
• This product creates a step-function improvement over the status quo.
• This business can become much larger than the initial wedge.
• This company can eventually own a durable control point.
• The current round unlocks specific milestones that materially increase value.

The difference between a weak pitch and a strong pitch is not the claims. Most decks make the same claims. The difference is evidence density.

A weak pitch says: “We are building an AI platform for a massive market.”

A strong pitch says: “We reduce claims review time from 47 minutes to 6 minutes for mid-market insurers, we are live with three design partners processing real claims, usage has grown week-over-week for eight weeks, our deployment sits inside the adjuster workflow, and the next product surface turns us from a point solution into the system of action.”

That second pitch gives us something to underwrite.

What we are actually trying to decide

At pre-seed and seed, we are not asking whether you have already de-risked everything. You have not. That is the point of the stage.

We are asking a harder question:

Is this a credible power-law candidate, and is this the right moment to invest?

Those are related questions, but they are not the same.

A company can have enormous upside but not be investable yet. A company can look very investable on current metrics but still lack venture-scale upside. A company can have brilliant founders and still be building into a market structure that will not support a fund-returning outcome. A company can have real revenue but poor revenue quality. A company can have strong usage but no path to margin control. A company can have a great wedge but no path to a control point.

We separate these questions deliberately:

1. Power-law potential

If everything works, can this become a company that matters at fund scale?

This is about upside. It includes market size, timing, category creation potential, founder-market fit, product differentiation, distribution, retention, margin quality, and durability.

2. Investment readiness

Is the company financeable now, at this stage, on these terms, with this evidence?

This is about the current check. It includes stage reality, proof level, traction quality, dependencies, regulatory exposure, valuation, burn, runway, sales cycle, and whether the round unlocks the right milestones.

The trap is confusing the two.

A big idea is not automatically an investable company. A company with revenue is not automatically a venture-scale company. A great team does not automatically overcome a broken thesis. And a fragile company is not automatically a bad investment if the upside is extraordinary and the fragility is the right kind.

Our job is to tell the difference.

What to send before we evaluate you

Missing information does not make us more optimistic. It forces conservative assumptions.

You do not need to have perfect data at pre-seed or seed. You do need to be clear about what is known, what is unknown, what is hypothesized, and what evidence you are collecting next.

Send the basics upfront.

Company and round

Include:

• Company name
• Website
• Headquarters or primary operating geography
• Stage
• Round type
• Round size
• Amount committed, if any
• Target valuation or valuation range, if applicable
• Previously raised capital
• Current cash position, if relevant
• Monthly burn
• Use of funds
• Milestones this round is meant to unlock

Do not just say, “This round gives us 18 months of runway.” Explain what happens during those 18 months.

Better:

“This round gets us to six enterprise deployments, $500K in contracted ARR, proof of 90-day retention across two cohorts, and enough usage data to demonstrate whether the workflow becomes daily.”

Founders and team

Include:

• Founder bios
• Prior startups, exits, or meaningful operating roles
• Domain expertise
• Technical depth
• AI, infrastructure, fintech, marketplace, or regulated-market expertise where relevant
• How long the team has worked together
• Who owns product
• Who owns engineering
• Who owns go-to-market
• Hiring plan for the next 12 months

Founder-market fit is not a credential contest. We care less about brand names than about earned insight, execution speed, recruiting ability, and evidence that you understand the non-obvious constraints of the market.

If the company is technical and there is no technical owner, say how that gap will be solved. If the company sells to a specialized buyer and nobody on the team understands that buyer, say how you are closing that gap. Silence looks worse than an honest plan.

Product, technical architecture, and velocity

Include:

• What is live today
• What is demoable
• What is production-grade versus prototype
• Shipping cadence
• Architecture overview
• AI model dependencies, if any
• Proprietary data, models, infrastructure, workflow, or integrations
• What is hard to copy
• What has improved materially over the last 30, 60, and 90 days

A demo is usually worth more than ten architecture slides. But the best demos are not feature tours. They prove a workflow change.

Show the old way. Show the new way. Show the delta.

Traction and customer proof

Include:

• Revenue
• Usage
• Growth rate
• Pipeline, clearly separated from signed revenue
• Production deployments
• Pilots
• Design partners
• LOIs
• Customer logos, with context
• Buyer title
• Contract status
• Sales cycle
• Deployment timeline
• Time to value

Do not blur categories.

A Fortune 500 pilot is not the same as production revenue. A design partner is not the same as a paying customer. An LOI is not the same as a signed contract. Pipeline is not revenue. GMV is not software revenue. Usage is not retention unless it repeats.

We are not offended by early-stage ambiguity. We are allergic to category confusion.

Retention, expansion, and reason to return

Include:

• Logo retention
• User retention
• Net revenue retention, if applicable
• Usage retention
• Cohort behavior
• Activation rate
• Expansion behavior
• Churn reasons
• What changed after churn
• Evidence that customers return without founder handholding

Retention is not only a SaaS metric. Every company needs a reason to return.

For consumer companies, that may be frequency and engagement. For marketplaces, it may be liquidity and repeat transactions. For infrastructure, it may be expanding usage. For enterprise workflow, it may be embedding into daily operations. For AI products, it may be whether the product becomes more useful as it is used (a data flywheel).

The question is simple: what makes this more than a one-time curiosity?

Market and budget

Include:

• Bottom-up market sizing
• Buyer
• User
• Budget owner
• Current spend
• Why the budget exists
• Why the problem is urgent
• Why the spend is non-discretionary or becoming non-discretionary
• Wedge market
• Expansion market
• Category-level upside

A giant TAM number is not a market argument. It is usually a weak substitute for one.

A strong market argument starts with a buyer and a budget line. Who pays? What do they pay for today? What breaks if they do nothing? What budget do you replace, expand, or create? How do you move from the wedge into the broader category?

Distribution and payback

Include:

• First acquisition channel
• Experiments already run
• Conversion rates
• Sales motion
• Sales cycle
• CAC, if known
• Payback logic
• Founder-led sales learnings
• Channel partnerships, if real
• Community or product-led loops, if real
• Why distribution gets easier over time

“We will do outbound” is not a distribution strategy. It is a default activity.

A stronger answer sounds like:

“We tested three channels. Founder-led outbound to VP Ops in logistics converted at 7% from first meeting to paid pilot. LinkedIn content produced awareness but weak intent. Partner referrals from implementation consultants generated fewer leads but 3x higher close rates. We are doubling down on the partner channel while using outbound to sharpen ICP.”

That shows learning, not theater.

Moat, control point, and durability

Include:

• Switching costs
• Workflow ownership
• System-of-record potential
• System-of-action potential
• Data rights
• Proprietary data loops
• Unique technical advantage
• Distribution advantage
• Embedded integrations
• Network effects, if any
• Why competitors cannot simply copy the feature and win

Early-stage moat is usually a trajectory, not a finished wall. That is fine. But you need to show the mechanism.

A feature is not a moat. A model wrapper is not a moat. An integration is not automatically a moat. A workflow wedge becomes interesting when it can compound into ownership of data, budget, behavior, distribution, or infrastructure.

The question we care about is not “what is hard today?” It is:

What gets harder for competitors as you scale?

Dependencies and regulatory exposure

Include:

• Platform dependencies
• Cloud dependencies
• Model dependencies
• Data provider dependencies
• Payment, banking, lending, or insurance dependencies
• Marketplace dependencies
• Channel dependencies
• Regulatory exposure
• Counterparty concentration
• Replacement timelines
• Redundancy status

If one API, one platform, one lender, one data provider, one marketplace, one channel, or one regulatory interpretation can break the company, name it.

We do not expect zero dependencies. We do expect you to know which dependencies matter.

A dependency map should answer:

• What can break us?
• How much revenue or usage depends on it?
• How long would it take to replace?
• What redundancy exists today?
• What redundancy is planned?
• What has already been tested?

“Planned redundancy” is not the same as live redundancy. Investors know the difference.

The first screen: eight questions that matter

We do not begin with vibes. We begin with gates.

A gate is a question that can cap the opportunity even if other parts of the company look strong. This is important because startups do not fail by averaging their strengths and weaknesses. They often fail because one unresolved issue breaks the thesis.

Here are the founder-facing versions of the questions we ask.

1. Are these the right founders for this market?

We look for founder-market fit, earned insight, intensity, execution speed, and the ability to recruit.

This does not mean every founder must have worked in the industry for 15 years. Sometimes an outsider sees what insiders normalized. But if you are an outsider, you need evidence that you have done the work: customer discovery, rapid product iteration, unusual insight, or a wedge that insiders missed.

What helps:

• You have lived the problem.
• You have built in the category before.
• You have sold to the buyer before.
• You have proprietary insight from prior work.
• You have unusual technical capacity.
• You have recruited people who should not be easy for you to recruit.
• You have learned quickly under pressure.

What hurts:

• The company feels like a market map exercise.
• The founders are chasing a trend without earned insight.
• No one owns the hard part of the business.
• The team has impressive logos but weak relevance.
• The founder story is mostly status, fundraising, or “AI is hot.”

Stated mission is cheap. Revealed commitment is expensive. We care about the latter.

2. Is the problem severe and is the timing real?

A severe problem is painful, frequent, expensive, urgent, or strategically important.

A timing argument explains why now is different.

Weak timing:

“The market is huge and AI is changing everything.”

Strong timing:

“Three things changed in the last 18 months: model accuracy crossed the threshold for this specific workflow, the buyer’s labor shortage created budget urgency, and new compliance requirements force every customer to revisit the process before next year.”

Timing is not just “the market is growing.” Timing is a wedge in reality.

We want to know:

• What changed?
• Why does that change matter?
• Why could this not have worked three years ago?
• Why might it be too late three years from now?
• Is regulation a tailwind, headwind, or forcing function?
• Are customers already changing behavior?

In fast-moving AI markets, being early is not automatically good. Early only matters if the head start converts into distribution, data, product depth, infrastructure ownership, or capital advantage. Otherwise, you may simply educate the market for the eventual winner.

3. Is there evidence of pull or a credible reason it will emerge?

At the earliest stages, pull can look different by company type.

For B2B software, it may be paid pilots, production deployments, expansion, or repeated usage.

For consumer, it may be retention, frequency, organic growth, or community intensity.

For marketplace, it may be liquidity, repeat transactions, supply-side retention, or demand-side urgency.

For infrastructure, it may be expanding usage, developer adoption, low churn, or integration depth.

For deep tech, it may be credible design partners, technical milestones, or non-obvious customer commitment.

We are careful with vanity traction.

Logos do not prove pull. Pilots do not prove pull. Waitlists do not prove pull. LOIs do not prove pull. Revenue does not prove pull if it is services-heavy, non-repeatable, or bought through founder heroics.

The question is:

What behavior shows customers want this to exist even when you are not pushing them?

4. Can this become large enough to matter?

Venture capital is not designed for merely good businesses. A solid $30M revenue company may be a great outcome for founders and employees, but it may not fit a seed fund’s return model.

We are trying to invest in companies that can produce outlier outcomes.

That requires:

• A large or rapidly expanding market
• A wedge into a much bigger category
• A business model that can scale
• Durable value capture
• A reason the winner can become very large
• A path to becoming a default system, network, or infrastructure layer

This does not mean you need to start with a giant horizontal product. Many great companies begin with a narrow wedge. But the wedge must open into something much larger.

We want to see both:

• Beachhead clarity: who urgently needs this first?
• Expansion logic: how does this become much bigger?

5. Is the wedge differentiated, and can it become a control point?

A wedge gets you into the customer.

A control point keeps you there and lets you expand.

Examples of potential control points include:

• Owning the workflow
• Owning the system of record
• Owning the system of action
• Owning the budget
• Owning proprietary data rights
• Owning distribution
• Owning a developer or partner ecosystem
• Owning infrastructure that others build on
• Owning a compliance layer that becomes hard to remove

A wedge without a path to control can still be a good product. It is often not a great venture investment.

This distinction matters especially in fast-moving AI markets. A useful feature can grow quickly and still get absorbed by a larger platform. A product can have early revenue and still lack a path to durable value capture. A team can execute well and still be trapped in the competent middle: good product, real customers, no strategic control point.

We want to know which shape you are building:

Distribution-first

You win by owning adoption. You spread quickly, create a usage loop, become the default surface, and earn the right to go deeper over time.

Evidence:

• Organic growth
• Bottom-up adoption
• Community pull
• Low CAC
• Viral or collaborative loops
• Strong user love
• Fast expansion inside accounts

Research or capital-intensive

You win by solving a hard technical problem that others cannot solve quickly.

Evidence:

• Exceptional technical team
• Proprietary research
• Hard technical milestones
• Capital intensity that creates a barrier
• A roadmap that compounds
• A product that depends on real technical depth, not prompt engineering

Infrastructure or platform-control

You win by becoming the layer others build on or depend on.

Evidence:

• Usage expands per account
• Customers integrate deeply
• Switching costs rise
• Developer or workflow dependency forms
• You control more margin over time
• You own the data, infrastructure, or execution layer that matters

“Both” can be powerful if both are evidenced. Distribution plus infrastructure depth is an exceptional shape. But claiming both without proof is just narrative inflation.

6. Are the economics durable?

Revenue quality matters more than revenue quantity.

A company can show impressive transaction volume, GMV, payment volume, usage, or pilot revenue without proving durable economics. We separate the types of revenue:

• Fixed subscription revenue
• Usage or consumption revenue
• Transaction or take-rate revenue
• Services revenue
• Implementation revenue
• Pass-through revenue
• Hardware or low-margin revenue
• Revenue dependent on third-party infrastructure

None of these are automatically bad. The question is whether the revenue is durable, high-margin, repeatable, controlled by the company, and capable of compounding.

Usage-based revenue, for example, can be excellent for infrastructure companies. The test is not “is it subscription?” The test is:

• Does usage expand inside accounts?
• Does the company control the margin path?
• Does deeper usage increase switching costs?
• Is the company building real infrastructure, or simply reselling someone else’s capability?

Consumption pricing can be high quality when usage grows and margin becomes owned. It can be low quality when usage is flat, churns, or depends permanently on expensive third-party inputs.

Do not pitch revenue volume as if it automatically equals software quality. Show the underlying economics.

7. Can the company survive the financing path?

Seed companies do not need to be default alive immediately. But the financing plan has to be coherent.

We look at:

• Current burn
• Post-raise runway
• Sales cycle
• Deployment time
• Time to value
• Expansion cycle
• Hiring plan
• Next-round milestones
• What happens if fundraising takes longer than expected
• Whether the company can shrink to a core wedge if needed

A common mistake is calculating runway without matching it to the go-to-market cycle.

If enterprise sales takes six months, deployment takes three months, proof of value takes three more months, and expansion takes another quarter, then 12 months of runway is not really 12 months of learning. It may be one sales cycle with no room for error.

We want to understand whether the round gives you enough time to produce the proof the next investor will need.

8. Does the product work today at the bar customers will apply?

This matters enormously for AI companies.

Many AI pitches depend on an unspoken assumption: “The models will get better.”

Sometimes that is a reasonable tailwind. Sometimes it is the entire thesis. Those are different.

If your product depends on a frontier capability that does not yet work at the required quality, we need to know:

• What exact capability is required?
• Does it work today?
• If not, how large is the gap?
• Who closes the gap: your team or an outside model provider?
• What happens if the capability does not improve for 18 months?
• What comparison class does the customer use?

A chatbot is judged against other software. A voice agent may be judged against a human. A video avatar may be judged against a human. A product that chooses a human comparison bar must clear a much higher threshold.

The key distinction is driver versus passenger.

A driver is closing the capability gap itself through owned models, owned infrastructure, proprietary data, specialized workflows, or deep technical execution.

A passenger is waiting for someone else’s roadmap.

Passengers can build interesting demos. Drivers can build companies.

How we classify your business

One of the fastest ways to lose investor trust is to pitch the business as one thing when the economics behave like another.

Many companies call themselves SaaS. Not all of them are SaaS.

You may actually be:

• AI-native software
• Vertical SaaS
• Infrastructure software
• Developer tooling
• Regulated workflow
• Payments or take-rate business
• Marketplace
• Data network
• Services-assisted software
• Deployment-heavy enterprise software
• Hardware-enabled software
• Usage-based infrastructure
• Capability-gated AI product
• Compliance or risk layer
• Workflow automation product
• System of record
• System of action

There is no shame in any of these categories. But each has different underwriting questions.

A marketplace needs liquidity.

A payments company needs margin and regulatory clarity.

An infrastructure company needs usage expansion and margin control.

A services-assisted product needs a credible path away from services drag.

A deployment-heavy enterprise product needs sales-cycle and implementation realism.

A regulated workflow company needs compliance depth and counterparty resilience.

An AI-native product needs capability readiness, differentiation, and model-cost awareness.

The wrong classification creates the wrong story. The wrong story creates the wrong valuation. The wrong valuation kills the financing path.

Be precise about what business you are actually building.

What “great” looks like by category

You do not need to be perfect across every dimension. At pre-seed and seed, no company is. But the best companies tend to be exceptional in a few load-bearing areas and honest about what remains unproven.

Founder-market fit

Great looks like:

• The founders have earned insight into the problem.
• The team owns the hardest parts of execution.
• The company is not a tourist in the market.
• The founders can recruit above their stage.
• The founders learn quickly from customer contact.
• The team can ship without outsourcing the core product.
• The founder story explains why this team sees what others missed.

Evidence to bring:

• Prior relevant work
• Customer discovery notes
• Product iterations
• Technical artifacts
• Hiring wins
• Advisory or customer validation
• Examples of fast learning
• Founder-led sales insights
• Domain-specific wedge insight

Common failure mode:

“We are ex-Google, ex-McKinsey, and the market is huge.”

That may be interesting. It is not enough. Pedigree helps only when it maps to the specific company-building problem.

Problem severity

Great looks like:

• The problem is painful and frequent.
• The current workflow is visibly broken.
• The buyer already spends money or time on the problem.
• The user has urgency.
• The cost of inaction is high.
• The problem is tied to revenue, cost, compliance, risk, labor scarcity, or strategic priority.

Evidence to bring:

• Customer quotes
• Current workflow screenshots
• Time-and-motion analysis
• Budget evidence
• Manual workarounds
• Churn from existing tools
• Compliance deadlines
• Labor cost data
• Error rates
• Revenue leakage
• Risk exposure

Common failure mode:

“This is inefficient.”

Most things are inefficient. The question is whether the inefficiency is painful enough to create behavior change.

Timing

Great looks like:

• There is a structural inflection.
• Customers are already changing behavior.
• New technology makes the product possible.
• New regulation creates urgency.
• A platform shift opens distribution.
• A labor, cost, or compliance shock changes budget priority.
• The old solution cannot adapt fast enough.

Evidence to bring:

• Recent customer behavior changes
• Budget shifts
• Regulatory timelines
• Technical performance improvements
• Platform changes
• Procurement changes
• Market pull that did not exist before
• Why incumbents are constrained

Common failure mode:

“AI makes this possible now.”

That is a slogan, not a timing argument. Be specific about what became possible, for whom, and why it matters commercially.

Product and technical differentiation

Great looks like:

• The product creates a 10x improvement in time, cost, accuracy, risk, or revenue.
• The demo maps to a real workflow.
• The architecture has defensible elements.
• The product improves with usage.
• The hard technical pieces are owned or being actively built.
• The team can explain why the product is hard to copy.
• Users would be upset if it disappeared.

Evidence to bring:

• Live demo
• Before-and-after workflow
• Benchmarks
• Latency, accuracy, or cost data
• Architecture diagram
• Model evaluation approach
• Integration depth
• Customer usage data
• User feedback
• Product velocity

Common failure mode:

“We use AI to automate the workflow.”

Which workflow? At what accuracy? With what latency? At what cost? With what failure handling? In whose system? Under whose permissions? With what data rights?

The difference between a demo and a company is hidden in those questions.

Capability readiness

Great looks like:

• The product works today for the beachhead customer.
• The core experience clears the customer’s comparison bar.
• Any remaining capability gap is owned by the company or clearly non-fatal.
• The product can survive if frontier models do not improve quickly.
• Model costs are understood.
• Failure modes are understood.
• The roadmap improves an already useful product, not a non-working promise.

Evidence to bring:

• Live customer usage
• Accuracy benchmarks
• Human evaluation
• Failure-case analysis
• Cost per task
• Latency data
• Customer tolerance thresholds
• Model fallback strategy
• Owned technical roadmap

Common failure mode:

“The models are improving fast.”

They are. But that does not answer whether your company captures the value.

Traction and growth

Great looks like:

• Growth is real and explainable.
• Customers are in production.
• Usage is expanding.
• Revenue quality is clear.
• Pipeline is tied to a repeatable motion.
• Design partners have budget and conversion intent.
• The founder understands what is driving growth.

Evidence to bring:

• Revenue by month
• Usage by cohort
• Pipeline by stage
• Signed contracts
• Deployment status
• Design partner conversion terms
• Customer concentration
• Expansion behavior
• Churn and churn reasons

Common failure mode:

“We have $2M in pipeline.”

Pipeline is useful context. It is not proof. Show stage, probability, source, buyer, next step, expected close date, and historical conversion.

Retention and reason to return

Great looks like:

• Customers keep using the product.
• Usage frequency matches the problem.
• Retention is not dependent on founder pushing.
• The product becomes more valuable over time.
• Expansion is visible or logically near.
• Churn is understood and acted on.

Evidence to bring:

• Cohort retention
• Repeat usage
• Account expansion
• NDR or usage expansion, if applicable
• Feature-level engagement
• Workflow dependency
• Renewal signals
• Churn notes
• Customer success learnings

Common failure mode:

“Users love it.”

Show behavior, not adjectives.

Distribution

Great looks like:

• The company has tested multiple acquisition paths.
• One channel is beginning to repeat.
• Sales motion matches ACV.
• CAC logic is grounded.
• Channel strategy is not theoretical.
• Distribution advantage compounds.
• The team knows who buys, who blocks, and who champions.

Evidence to bring:

• Funnel metrics
• Outreach conversion
• Source-by-source pipeline
• Sales-cycle data
• CAC assumptions
• Payback logic
• Partner contribution
• Community growth
• Product-led loops
• Expansion inside accounts

Common failure mode:

“We will hire sales after the round.”

Hiring sales is not a go-to-market strategy. It is an expense line.

Market size and expansion

Great looks like:

• The wedge is narrow enough to win.
• The category is large enough to matter.
• Bottom-up sizing is credible.
• Buyer and budget are identified.
• Expansion is logical.
• The company can become a default layer if it wins.

Evidence to bring:

• Number of target buyers
• Budget per buyer
• Current spend
• Initial ACV
• Expansion ACV
• Adjacent buyer segments
• Category map
• Competitive replacement path
• Why the winner can be large

Common failure mode:

“This is a $100B market.”

A large market with no access path is not a strategy. It is scenery.

Moat and control point

Great looks like:

• The product gets harder to remove over time.
• Data advantage grows with usage.
• Workflow dependency deepens.
• Integrations become meaningful switching costs.
• The company can own a system of record or action.
• Distribution advantage compounds.
• The company captures more margin over time.
• Competitors can copy features but not the accumulated position.

Evidence to bring:

• Integration depth
• Data rights
• Workflow maps
• Admin or compliance dependency
• User collaboration loops
• Developer ecosystem
• Proprietary data loops
• Switching-cost examples
• Expansion from wedge into platform

Common failure mode:

“Our moat is that we move fast.”

Speed matters. Speed alone is not a moat. Plenty of failed startups shipped quickly. Velocity matters when it compounds into data, distribution, switching costs, or infrastructure ownership.

Revenue quality and business model

Great looks like:

• Revenue is repeatable.
• Gross margin can become attractive.
• Expansion is plausible.
• The company controls the customer relationship.
• The company controls or can control key cost inputs.
• Services do not dominate the long-term model.
• Usage-based revenue expands and improves margin over time.
• Pricing maps to value.

Evidence to bring:

• Revenue by type
• Gross margin
• Model or infrastructure costs
• Services percentage
• Implementation cost
• Usage expansion
• Contract terms
• Pricing tests
• Willingness-to-pay evidence
• Margin roadmap

Common failure mode:

“We processed $10M.”

Processed volume is not revenue. Revenue is not margin. Margin is not control. Show the whole chain.

Capital efficiency and milestones

Great looks like:

• Burn matches learning velocity.
• The round funds specific proof points.
• Hiring plan is tied to milestones.
• Runway matches sales and deployment cycles.
• The company has a credible plan if fundraising is delayed.
• The next round can be raised on measurable progress, not narrative.

Evidence to bring:

• Current burn
• Post-raise burn
• Hiring plan
• Runway
• Milestone schedule
• Sales-cycle assumptions
• Deployment timing
• Financial model
• Downside plan
• Shrink-to-core plan

Common failure mode:

“This gives us 18 months.”

To do what? With what proof at the end? For which next investor? At what valuation logic?

Special guidance for AI companies

AI has made it easier to build impressive demos and harder to identify durable companies.

We like AI-native companies. But we do not give credit for AI theater.

If AI is core to the pitch, answer these questions directly.

What is the actual capability?

Do not say “agents,” “automation,” or “AI workforce” and stop there.

Say what the product must do:

• Read unstructured documents with a specific accuracy threshold
• Complete a workflow across systems
• Make real-time recommendations
• Generate voice at human-comparable latency
• Produce compliant outputs
• Reason over proprietary data
• Take actions with acceptable error rates
• Replace manual QA
• Reduce time-to-resolution
• Handle edge cases in production

What is the comparison bar?

A back-office summarization tool is judged differently from a customer-facing voice agent.

If your product replaces a human, customers judge it against a human. If your product augments a workflow, customers judge it against the current workflow. If your product sells to developers, they judge it against speed, control, reliability, and trust.

Name the bar. Then show you clear it.

Are you a driver or a passenger?

If your product depends on OpenAI, Anthropic, Google, Meta, Mistral, or another model provider improving a capability, say so. That may be acceptable, but it is a dependency.

If you are building your own infrastructure, fine-tuning, owning inference, creating proprietary data loops, or narrowing the problem so you can outperform general models, show that.

The question is not whether foundation models improve. They will. The question is whether your company captures value as they improve.

What happens if the frontier stalls?

For the next 18 months, assume the specific capability you need does not get much better.

Does the product still work?

If yes, we want to know why.

If no, then the pitch is not just a startup pitch. It is a bet on someone else’s research roadmap.

That can be a dangerous place to be.

Special guidance for infrastructure and usage-based companies

Usage-based pricing is not a weakness by itself. For infrastructure, it is often the right model.

But usage-based revenue must pass a different quality test.

We want to see:

• Existing customers using more over time
• Clear margin path
• Cost inputs that improve with scale
• Infrastructure ownership or credible movement toward it
• Deep integration
• Developer or workflow dependency
• Evidence that usage creates switching costs

Bad usage revenue looks like rented COGS and shallow customer dependence.

Good usage revenue looks like expanding customer consumption on infrastructure you increasingly control.

The difference matters.

If you are an infrastructure company, do not pretend to be classic SaaS. Explain why the infrastructure model is better. Show usage expansion. Show margin control. Show why customers build around you.

Special guidance for marketplaces and network-dependent businesses

For marketplaces and network-dependent businesses, we care about liquidity, control, and take-rate durability.

Bring:

• Supply growth
• Demand growth
• Match rate
• Fill rate
• Repeat rate
• Contribution margin
• Take rate
• CAC by side
• Payback by side
• Disintermediation risk
• Concentration risk
• Network effects, if real
• Why the marketplace gets stronger with scale

Marketplace decks often overstate network effects. A network effect exists when each incremental participant makes the product more valuable for other participants. Merely having supply and demand is not enough.

We also want to know who can compress you:

• Suppliers
• Buyers
• Platforms
• Payment processors
• Regulators
• Incumbents
• Aggregators

If the network works, why do you keep the margin?

Special guidance for fintech, regulated workflow, and counterparty-dependent companies

In fintech and regulated markets, the startup often depends on banks, lenders, payment rails, insurers, data providers, regulators, or compliance interpretations.

That does not make the business bad. It does make hand-waving dangerous.

Bring:

• Regulatory map
• Licensed versus unlicensed activities
• Partner bank or counterparty dependencies
• Funds flow
• Data flow
• Compliance ownership
• Failure modes
• Concentration exposure
• Redundancy plan
• Legal or regulatory counsel, if relevant
• Unit economics after partner costs
• What happens if one counterparty exits

Regulated companies can become extremely valuable because trust and compliance create barriers. But those same barriers can kill a company that scales on a fragile foundation.

We need to know which version you are building.

How to structure a high-conviction pitch deck

A good deck reduces ambiguity. A great deck makes the investment case easy to reconstruct after the meeting.

Use this sequence.

1. One-line company description

Say what you do in plain English.

Bad:

“We are redefining the future of enterprise intelligence.”

Better:

“We automate prior authorization for specialty clinics, reducing manual review time by 80% while integrating into the systems nurses already use.”

2. Vision

What category are you trying to own?

This should be ambitious but not vague. The best vision slides connect the wedge to the eventual platform.

3. Problem

Show the pain.

Include:

• Who has the problem
• How often it happens
• What it costs
• Why current tools fail
• What customers do today
• Why the problem is urgent now

4. Why now

Explain the inflection.

Include:

• Technology shift
• Market shift
• Regulatory shift
• Labor shift
• Buyer behavior shift
• Platform shift
• Cost curve shift

Be specific.

5. Product

Show the product early.

A demo-first deck is often stronger than a narrative-first deck. Investors want to see the thing.

6. Workflow before and after

This is one of the most underused slides.

Show:

• Old workflow
• New workflow
• Time saved
• Cost reduced
• Error reduced
• Revenue gained
• Risk removed
• User behavior changed

7. Technical differentiation

Explain what is hard.

Include:

• Architecture
• Model strategy
• Data advantage
• Integrations
• Infrastructure
• Security
• Compliance
• Latency
• Accuracy
• Cost advantage
• Proprietary workflow logic

Do not drown the deck in technical detail. But do show enough for us to believe there is substance.

8. Customer and buyer

Separate user, buyer, budget owner, and champion.

Many enterprise startups fail because the user loves the product but the buyer does not own budget. Make the buying path explicit.

9. Traction

Show real proof.

Break out:

• Signed revenue
• Production deployments
• Paid pilots
• Unpaid pilots
• Design partners
• Pipeline
• LOIs
• Waitlist

Do not merge them into one big number.

10. Retention and usage

Show whether customers come back.

Include:

• Cohorts
• Frequency
• Expansion
• Activation
• Repeat usage
• Churn
• Churn reasons
• Lessons from churn

If it is too early for retention, show the best proxy and say why it matters.

11. Business model

Explain how you make money.

Include:

• Pricing
• ACV
• Gross margin
• Services component
• Usage costs
• Expansion logic
• Contract structure
• Expected margin at scale

If you are usage-based, show why usage expands and how margin improves.

12. Distribution

Show how you acquire customers.

Include:

• Channels tested
• Conversion metrics
• Sales cycle
• CAC logic
• Payback logic
• Repeatability
• Why this channel scales
• What you learned from failed channels

13. Market

Use bottom-up logic.

Include:

• Number of target customers
• Budget per customer
• Initial wedge
• Expansion path
• Adjacent markets
• Why the winner can become large

Top-down market slides are fine as context. They should not be the main argument.

14. Competition

Do not use a fake 2x2.

Explain:

• What customers use today
• Why incumbents have not solved it
• Why startups may attack it
• Why you win the wedge
• Why you can expand
• What happens if a platform copies the feature

The best competition slides show strategic understanding, not visual positioning.

15. Moat and control point

Explain what compounds.

Include:

• Data
• Workflow ownership
• Switching costs
• Integrations
• System-of-record path
• System-of-action path
• Developer or partner ecosystem
• Infrastructure ownership
• Distribution loops

16. Team

Explain why this team wins.

Tie the team directly to the company-building problem.

Do not just list logos. Explain relevance.

17. Risks and mitigations

This slide builds trust when done well.

Include:

• Technical risk
• Market risk
• Sales risk
• Regulatory risk
• Dependency risk
• Model risk
• Financing risk
• Competitive risk

For each, say what you have done and what remains.

A founder who can name the real risks is easier to back than a founder pretending none exist.

18. Round and milestones

End with the financing plan.

Include:

• Amount raising
• Instrument
• Valuation or valuation range, if applicable
• Committed capital
• Use of funds
• Hiring plan
• Milestones unlocked
• Expected runway
• Next-round proof points

The round slide should answer: why is this amount of capital the right amount right now?

Weak versus strong: examples

Team

Weak:

“Our team has experience at top companies.”

Strong:

“Our CTO built low-latency inference systems at scale, our CEO sold workflow software into this buyer for six years, and our founding engineer previously maintained the open-source library that this customer segment already uses.”

Problem

Weak:

“The process is broken and inefficient.”

Strong:

“Revenue operations teams spend 12–18 hours per week reconciling billing exceptions across Salesforce, Stripe, and NetSuite. Errors delay collections by an average of 21 days and create direct cash-flow impact.”

Timing

Weak:

“AI is finally good enough.”

Strong:

“Three changes make this possible now: LLM extraction accuracy crossed the threshold for semi-structured invoices, finance teams are under pressure to reduce headcount growth, and CFOs are consolidating point tools after two years of SaaS sprawl.”

Product

Weak:

“Our AI agent automates the workflow.”

Strong:

“The agent completes the first-pass review, flags exceptions, writes back to the system of record, and asks for human approval only on low-confidence cases. In live customer tests, it handled 63% of cases without intervention.”

Traction

Weak:

“We have five enterprise logos.”

Strong:

“Two are in paid production, one is a paid pilot, and two are unpaid design partners. The paid production customers process 14,000 tasks per month through the product, and one expanded from one team to three teams in the first 60 days.”

Retention

Weak:

“Customers love us.”

Strong:

“Week-eight account retention is 92%, retained accounts increased usage 34% month-over-month, and churned users cited missing integrations that we have since shipped.”

TAM

Weak:

“This is a $50B market.”

Strong:

“There are 18,000 target buyers in our initial segment. Based on current spend and our tested pricing, the wedge market is $1.4B. Expansion into adjacent workflows increases ACV by 3–5x, creating a credible path to a $10B+ category.”

Distribution

Weak:

“We will scale with outbound and partnerships.”

Strong:

“Outbound to VP Operations converts poorly unless triggered by a compliance deadline. Partner referrals from implementation firms convert 4x better and shorten sales cycles by 40%. We signed two referral partners and sourced three pilots from that channel.”

Moat

Weak:

“Our moat is proprietary AI.”

Strong:

“Each deployment creates labeled exception data, customer-specific workflow logic, and deep integrations into approval systems. After 90 days, the product becomes the routing layer for exceptions, making replacement operationally painful.”

Dependencies

Weak:

“We use standard third-party providers.”

Strong:

“Today, 80% of inference runs through one model provider. We have implemented fallback routing for two lower-complexity workflows, but not for the highest-value workflow. Reducing this dependency is one of the three core milestones of this round.”

The fastest ways founders score themselves down

1. Distribution hand-waving

“We will do outbound” is not enough.

Show actual experiments, conversion rates, sales-cycle data, channel learning, and why the motion can scale.

2. Top-down TAM only

A big number without buyer-budget logic is weak.

Show who pays, what budget you attach to, and how the wedge expands.

3. AI narrative without AI reality

“AI-powered” does not mean differentiated.

Show what works today, what is proprietary, what the comparison bar is, and whether you are a driver or passenger on the required capability.

4. Logos without context

Logos can mislead.

Separate production, paid pilot, unpaid pilot, design partner, LOI, and pipeline.

5. Confusing volume with revenue quality

GMV, TPV, API calls, messages, documents processed, and transactions can be useful. They are not the same as durable, high-margin revenue.

Explain margin, control, and expansion.

6. Moat confusion

Features are not moats. Integrations are not automatically moats. Data is not automatically a moat.

Show what compounds.

7. Dependency blindness

One platform, one lender, one data provider, one cloud service, one model provider, one channel, or one regulatory interpretation can be enough to cap the company.

Name the dependency and show mitigation.

8. Stage mismatch

A large round or aggressive valuation before the right proof can make a good company harder to finance.

Round size should match the evidence you have and the milestones you can reach.

9. Metrics fog

Dashboards are not explanations.

Tell us what changed, why it changed, which metrics matter, what is leading versus lagging, and what you learned.

10. “The model will improve” as a business plan

Model improvement may help. It may also commoditize your product.

Explain how your company captures value as the underlying capability improves.

What decisions mean

We try to be direct about outcomes.

Fast yes

This happens when upside is large, the team is credible, the wedge is differentiated, proof is strong for stage, and there are no unresolved thesis-breaking risks.

A fast yes does not require every metric to be mature. It requires enough evidence that the remaining risk is worth taking.

Priority meeting

This happens when the upside could be exceptional but a small number of questions need to be resolved in conversation.

Examples:

• Technical differentiation is promising but needs deeper review.
• Customer pull appears real but needs context.
• The company may be genuine infrastructure but we need to understand margin and control.
• The AI capability appears to work for a beachhead but needs validation.
• The founder-market fit may be exceptional but is not obvious from the deck.

Re-engage when proof arrives

This is not a polite no. It means the company may become investable if specific evidence appears.

Examples:

• Convert design partners into production customers.
• Prove repeat usage.
• Show expansion inside accounts.
• Reduce dependency concentration.
• Ship the technical milestone.
• Clarify budget ownership.
• Bring valuation in line with proof.
• Demonstrate a repeatable acquisition channel.

The best founders treat this as a milestone list.

Pass

We pass when the upside is not venture-scale, the proof is too weak for the stage, the economics do not support the narrative, the business is misclassified, or a critical risk breaks the thesis.

A pass is not always a judgment on founder quality. Sometimes it is market structure. Sometimes it is timing. Sometimes it is financing risk. Sometimes it is simply not a fit for our fund model.

How to build your own pre-pitch scorecard

Before sending a deck, memo, and other materials, complete this exercise.

For each category, write one sentence that states your claim. Then list the evidence you can show in 60 seconds.

Team

Claim:

Why are you unusually equipped to win?

Evidence:

• Relevant experience
• Earned insight
• Technical depth
• Customer access
• Recruiting ability
• Execution speed

Problem

Claim:

What severe problem are you solving?

Evidence:

• Frequency
• Cost
• Urgency
• Current workaround
• Budget
• Customer quotes

Timing

Claim:

What changed that makes this possible or necessary now?

Evidence:

• Technology shift
• Regulation
• Market behavior
• Labor or cost pressure
• Platform change
• Budget movement

Product

Claim:

What is live, and why is it meaningfully better?

Evidence:

• Demo
• Workflow delta
• Accuracy
• Latency
• Cost
• User feedback
• Production usage

Capability readiness

Claim:

Does the product work today at the bar customers apply?

Evidence:

• Benchmarks
• Live usage
• Failure cases
• Model dependency map
• Cost per action
• Fallback plan

Traction

Claim:

What proof shows customers want this?

Evidence:

• Revenue
• Production usage
• Paid pilots
• Design partners
• Growth
• Pipeline quality

Retention

Claim:

Why will customers keep using it?

Evidence:

• Cohorts
• Repeat usage
• Expansion
• Churn analysis
• Workflow dependency

Distribution

Claim:

How do you acquire customers repeatably?

Evidence:

• Channel tests
• Conversion rates
• CAC logic
• Sales cycle
• Partner results
• Product-led loops

Market

Claim:

How does the wedge become a large company?

Evidence:

• Bottom-up sizing
• Buyer count
• Budget
• ACV
• Expansion path
• Category map

Moat

Claim:

What compounds as you scale?

Evidence:

• Data rights
• Switching costs
• Integrations
• Workflow ownership
• Infrastructure ownership
• Network effects

Business model

Claim:

Why does this become durable, high-quality revenue?

Evidence:

• Pricing
• Gross margin
• Revenue mix
• Usage expansion
• Services percentage
• Margin roadmap

Dependencies

Claim:

What can break the company, and what are you doing about it?

Evidence:

• Top dependencies
• Concentration
• Replacement time
• Redundancy
• Regulatory plan

Financing plan

Claim:

What does this round unlock?

Evidence:

• Use of funds
• Hiring plan
• Runway
• Milestones
• Next-round proof
• Downside plan

If you cannot support a claim with evidence, do not hide it. Reframe it as a hypothesis and explain the experiment that will prove or disprove it.

That builds trust.

The best founders make the investor’s job easier

The strongest founders are not the ones who pretend the company is already de-risked. They are the ones who know which risks matter.

They can say:

• “This is proven.”
• “This is partially proven.”
• “This is still a hypothesis.”
• “This is the next experiment.”
• “This is the risk that scares us most.”
• “This is what would change our mind.”
• “This is what we learned when customers did not buy.”
• “This is why the upside is still worth it.”

That is how venture conviction forms.

Not from perfect companies. Not from perfect decks. From founders who combine ambition with precision.

What we want to believe after your pitch

By the end of a strong pitch, we should be able to say:

1. This could be a very large company if it works.
2. These founders have a credible right to win.
3. The problem is severe enough to drive behavior change.
4. The product is not just interesting; it is meaningfully better.
5. The timing creates urgency.
6. The wedge can expand into a bigger control point.
7. The business model can produce durable value capture.
8. The company understands its dependencies and risks.
9. The round unlocks milestones that matter.
10. The remaining uncertainty is the kind of risk venture capital is supposed to take.

That last point matters.

We are not trying to eliminate risk. We are trying to distinguish good risk from bad risk.

Good risk is technical ambition, market creation, rapid scaling, or an insight that may look strange before it works.

Bad risk is unclear ownership, weak customer pull, misclassified economics, fragile dependencies, fake traction, or a thesis that depends entirely on someone else’s roadmap.

Bring us the first kind.

Closing

We are looking for founders building companies that can become much larger than they look today.

That means we will sometimes lean into companies that look early, strange, or fragile by conventional standards. But we will not confuse narrative with evidence. We will not treat logos as retention. We will not treat volume as margin. We will not treat an integration as a moat. We will not treat a demo as capability readiness. We will not treat pedigree as proof.

The best pitches do not try to game the investor.

They make the truth legible.

They show what is working, what is not yet proven, why the upside is enormous, and why this team is the one to pursue it.

Look forward to seeing what you are building.

Pitch us here: https://tr.ee/pitch-us