Phronel
v1.0by Caplia Labs

MeetPhronel.

Private AI to read early-stage companies
the way you do, but with an edge.

You're probablyalready using AI.

To review documents. To analyse businesses. To make sense of complex information. It's fast, it's everywhere, and it mostly works.

But.

General-purpose AI wasn't built to understand your criteria, protect sensitive company data or deliver consistent, evidence-backed decisions.

Where general AI falls short.

01It's a generalist.It knows a little about everything. It doesn't know what a good early-stage company actually looks like.
02It doesn't know how you decide.No thesis. No criteria. Ask twice, get two answers, and no way to check its working.
03It isn't built for secrets.Sensitive company data, your own decision logic and internal notes should never become training materials.
Our mission

Find the needle.Not a copy of the last one.

The best founder in your pipeline rarely looks like the last one you backed. Tools that learn only from what you liked before narrow the funnel to more of the same, and quietly bake in bias. Phronel is built the other way round: surface the best founders on evidence, then give your team back the hours to judge what only people can.

01Evidence, not resemblancePhronel scores readiness against the outcomes of companies that have seen traction, across the market. Not against how closely a company resembles your past deals.
02Fit informs. It never filters.Your criteria sits next to readiness, not on top of it. A strong company that looks nothing like your portfolio is flagged for a human look, not quietly screened out.
03What they've done, not who they resemblePhronel reads traction, capital and execution the same way for every company, and flags spikes in sectors, places and founder profiles you've never backed.
04More time spent on peoplePhronel does the reading, so your team spends its hours with founders: how they think, decide and lead under pressure. The judgement no model should make for you.

Your criteria guide the search. They never decide which founders you get to see.

What Phronel is

Private AI, built forearly-stage.

Four parts, one intelligence. Our own frameworks and tools, working as one so you make better calls, faster. Developed by the brightest minds in early-stage company evaluation.

01 · FrameworkCRIThe Caplia Readiness Index. How ready is a company, and what should you ask?
02 · ToolP1515 traction signals, flagged the moment one spikes.
03 · ModelReasoning layerFine-tuned for early-stage, applied to your own criteria.
04 · InfrastructurePrivateYour decision-logic and company data stay yours. Always.
01 / CRI · Our readiness framework

Know how ready a companyreally is. Instantly.

The Caplia Readiness Index (CRI) scores 28 signals across five areas, and surfaces the fundamental questions to ask before you're in the room. One common readiness language, for every company you see.

3.9xIt's robust. Companies scoring 600+ on the CRI are in a fundamentally better state, statistically 3.9x more likely to raise capital within 12-months than those scoring below 300.
The market changes daily. That's why the CRI is versioned. We take the outcomes of companies with proven traction and re-weight the framework so it's in-line with what's happening in market.Built by Caplia Labs, in research collaboration with Oxford Saïd Business School.
Team6 signals
Product & Competition6 signals
Market5 signals
Traction6 signals
Foundational Readiness5 signals
Versioned · Re-weighted on real capital outcomes
02 / P15 · Our traction tool

Spot tractionbefore everyone else.

P15 watches 15 traction signals for every company you follow. When one spikes, Phronel tells you, in plain language, what changed, the evidence behind it and why it matters.

15Traction signals, always onFifteen distinct signals tracked for every company in your pipeline or portfolio, not just the ones you happen to check.
↑Flagged when it spikesThe moment a signal moves, Phronel surfaces it. No dashboards to trawl, no updates missed.
“ ”Explained, not just scoredEvery spike comes back to you qualitatively: what happened, the evidence behind it, and why it matters to your thesis.
03 / Reasoning layer

It thinks inearly-stage.

A reasoning layer for the early-stage company ecosystem, built with academics and leading institutions, with infinite knowledge (not generic web knowledge). It follows the reasoning experienced investors, banks and early-stage programme leads already use, and applies your criteria, not someone else's.

Every finding traces back to its source.
Strengths, risks and next questions, consistently.
Consistent foundations. Your judgement makes the decision.
General AIKnows a bit about everything
Phronel reasoning layerFine-tuned for early-stage
Your thesis and criteriaHow you decide
↓
A clear answer you can checkStrengths · Risks · Next questions
04 / Private

Private.Not private-ish.

Public AI tools were never built for founder data, term sheets, internal memos or committee debate. Phronel is.

Hosted on our own hardware.Phronel runs on Caplia's own hardware, inside an environment we control and secure.
Never used for training.Your thesis, criteria, prompts and founder information are never used to train or improve Phronel. Your edge stays yours.
Or run it on your own servers.Deploy Phronel inside your own perimeter, so not even Caplia can see your data.
Reporting

A credit score forearly-stage companies.

Lenders have a common score. Early-stage companies never did, and our intelligence becomes that standard. Every Phronel assessment produces an evidenced report on one consistent scale, written in the language your committee, credit team, panel or board already uses. Share it with confidence, and keep it on file as part of your audit trail.

Every report includes
Evidenced. Versioned. Ready to share.
Score0 to 1,000, with the five areas behind it
EvidenceEvery finding linked to its source
CriteriaYour framework and terminology, applied
QuestionsRisks and what to ask next
RecordTimestamp, framework version and reviewer
SharePDF, link, or straight into your CRM
01One score, one scaleEvery company is scored from 0 to 1,000 on the same framework, so results compare across teams, programmes and years.
02Every finding evidencedEach point links back to the document, data or conversation it came from. Nothing to take on trust.
03Speaks your languageFramed for the room it is going into, whether that is an investment committee, a credit panel, a grant board or a partner meeting.
04Built for the audit trailTimestamped and versioned. See what was assessed, when, against which framework, and why the decision was made.
Example · Venture capital

We deliver impact.Matched to a partner's judgement.

An early-stage fund gave Phronel a first set of companies it had already judged. Phronel's scores lined up with the partner's own calls, and became decision bands the team could stand behind: the clear weak tail passed automatically, and anything borderline still goes to a human.

77–100Strong fit
60–76Credible fit
50–59Manual review, always a human
0–49Automatic pass
Decision bands set from the fund's own labels, and recalibrated as more labels come in.
83%alignment with the partner's own calls on the first calibration set
0thesis documents or sensitive fund materials used to train a model
50%+reduction in time spent screening, compared with human review alone
Two ways in

Works where youalready work.

Use Phronel in our operational platform, configured to you, or within the tools you already use. Same intelligence, same privacy.

01 / Our platformVia Caplia's platform, your configurationYour own Caplia workspace, set up with your thesis, criteria, scoring weights and team roles. Phronel runs across intake, pipeline, portfolio and programme views.
IntakePipelinePortfolioProgrammes
02 / Your toolsWithin your existing toolsChrome, Slack and Microsoft Teams, your CRM via REST API, or your own agents over MCP. Your prompts stay private, never sent to OpenAI, Anthropic or Gemini.
ChromeSlackMicrosoft TeamsREST APIMCP
Pricing · agentic access

We sell decisions, not seats.

Pay as you go, or choose a monthly plan and draw on it from any surface. Unlimited users on every plan. The Caplia Enterprise Platform is live today and priced separately.

● Live todayThe Caplia platform is live today.Intake, CRI scoring, thesis matching and portfolio views, ready for your team now.
1
Analysis creditsOne credit is one company analysis: a full report on a company, read like a credit report. Its CRI, P15 and a snapshot of the company against your decision logic, with the inconsistencies, open questions and flags to raise before your next meeting. Runs wherever you already work: in Chrome, Slack or Teams, through the API, or over MCP inside your own tools.
T
Research tokensIncluded with every plan, pay as you go too, for the research your team would otherwise do in Claude or ChatGPT: questions, market and sector research, thesis work, memo drafting and deep research across many sources. Use them in the Caplia platform or in your own workflows, all kept private.
How far research tokens goTokens measure how much Phronel reads and writes. Longer, source-heavy work uses more.
~10KtokensA research questionA question answered with a few sources, like a typical chat.
~200KtokensA deep research reportDozens of sources read and synthesised into a full report.
Plans
Pay as you go
$14per company analysis
No commitment pay only for the analyses you run1M research tokens a month, about 100 questions or 5 deep research reports, in the platform or your own workflowsUnlimited users on one shared account$14.00 per company analysis
Angels, scouts and smaller teams analysing a handful of companies, or anyone trying Phronel before committing to a plan.
40 analyses / month
$299per month
40 analysis credits a month2M research tokens a month, about 200 questions or 10 deep research reportsUnlimited users on one shared pool$7.50 per extra analysis if you go over your allowance, same rate as the plan
Family offices investing directly, RIAs, and Caplia Platform customers adding intelligence to a programme.
200 analyses / month
$1,199per month
200 analysis credits a month10M research tokens a month, about 1,000 questions or 50 deep research reportsUnlimited users on one shared pool$6.00 per extra analysis if you go over your allowance, same rate as the plan
Venture funds, corporate venture units, consultancies and corporate finance advisory.
800 analyses / month
$3,999per month
800 analysis credits a month50M research tokens a month, about 5,000 questions or 250 deep research reportsUnlimited users on one shared pool$5.00 per extra analysis if you go over your allowance, same rate as the plan
High-volume teams, and software, data and workflow vendors using Phronel at scale.
Going over your plan?Keep going. Every analysis beyond your monthly allowance is charged at your plan's per-analysis rate: $7.50, $6.00 or $5.00. No cut-offs, no penalty pricing.
Multiplayer by default: no seats, no per-user licences. Your whole team works from one shared pool. The larger the plan, the lower the price per analysis: $14.00 down to $5.00. All prices in USD, billed monthly.
Working out the plan you need
1creditfor every new company you analyse
0.1creditper company you monitor continuously, each month
Estimate your monthly plan
You need about 40 analysis credits a month→ 40-credit plan · $299/mo
An angel analysing 10 companies a month10/mo → Pay as you go · $140
A fund analysing 25 new companies a month and monitoring 150 continuously40/mo → 40-credit plan
A programme screening 150 applicants a month and monitoring 500 alumni200/mo → 200-credit plan
A fund on the 40-credit plan that analyses 50 companies one month50/mo → $299 + 10 × $7.50
Continuous monitoringA point-in-time analysis starts ageing the moment it's written. Put a company on monitoring and it stays current instead. It draws from the same monthly pool, so you can move between one-off analysis and continuous coverage without changing your plan.
0.1credit per company per month
A fund monitoring 100 portfolio companies spends 10 credits a month doing it.
Putting Phronel inside your own productIf Phronel is going to reach people beyond your own team, the rights tier depends on how far it travels. Each is a multiplier on the credit price, with a minimum monthly pool.
RightsMinimum poolMultiplierFrom
EmbeddedPhronel runs inside your product, for your own users800 credits / mo1.25×$5,000 / mo
OEM & white-labelPresented under your brand rather than ours2,000 credits / mo1.50×$13,500 / mo
Redistribution & resaleYou sell Phronel intelligence on to your own customers4,000 credits / mo2.00×$32,000 / mo
What isn't a creditAPI access on a Caplia Platform plan connects your systems to your own programme data: applications, cohorts, companies you already work with. That's included in the Platform plan and costs no credits. Adding teammates never costs anything either.
One rule, everywhereEvery new company analysis, the full report on one company, costs one analysis credit, whether you trigger it from the Platform, the API, MCP or the browser extension. General questions and deep research draw on your plan's research tokens.
Not sure which plan?Tell us how many companies you expect to analyse each month and how many you'd want kept current, and we'll size it with you. Start on pay as you go and move up or down a plan at any time.
Size your plan with us
Phronel by Caplia · pricing current as of September 2026 · All prices in USD, billed monthly · Token equivalents are typical, actual usage varies with sources and length
Phronel vs frontier models

The frontier ofearly-stage intelligence.

Frontier models are brilliant generalists. Phronel is the specialist: private intelligence built for one job, reading early-stage companies the way experienced investors, banks and programme teams do.

CapabilityPhronelChatGPTClaudeGeminiGrokLlama
Leading empirically backed readiness scoring (CRI™)
Purpose-built for early-stage company evaluation
Calibrated on real capital outcomes, in research collaboration with Oxford Saïd Business School
Same company, same score, on one 0 to 1,000 scale
15 traction signals tracked, flagged the moment one spikes (P15)
Applies your thesis and criteria the same way, every time
Every finding linked to its source
Audit-ready reports: timestamped, versioned, evidenced
Your thesis, prompts and founder data never used for training
Runs on dedicated hardware, or on your own servers
Works in your tools: Slack, Teams, Chrome, your CRM and MCP
Built in Partial, or depends on plan and setup Not offered
Based on each provider's publicly available, general-purpose offerings, October 2026. Enterprise terms vary. ChatGPT, Claude, Gemini, Grok and Llama are trademarks of their respective owners.
Phronel by Caplia

The secure, private AI for early-stage company analysis.

UpdatesGet Caplia updates, product releases and venture insights straight to your inbox.
© 2026 Caplia Technologies Ltd. All rights reserved.
UpdatesGet Caplia updates, product releases and venture insights straight to your inbox.
© 2026 Caplia Technologies Ltd. All rights reserved.