Insurance / Fintech · Deep dive

Taktile

A no-code decision engine that lets banks, lenders and insurers build, backtest and run automated risk decisions — credit, fraud, KYC and underwriting — then rebrands the whole thing as an 'agentic' decision platform.

emerging

The question that decides it: Taktile's bet is that risk and underwriting teams need a dedicated, vendor-owned layer to assemble data, rules and models into governed decision flows. Does that horizontal layer stay essential once foundation-model agents can let a carrier or lender wire their own bureaus, KYC feeds and scorecards into an in-house flow — or does Taktile's edge collapse to the parts (audit trail, model-risk governance, integration maintenance) that a regulated buyer would rather not own but also will not pay a premium for?

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Taktile sells the software layer where a financial institution’s automated yes/no decisions actually get made — whether to approve a loan, onboard an account, flag a transaction as fraud, or accept an insurance risk. Founded in Berlin in 2020 by two ex-QuantCo data scientists who met at Harvard, it packages data integrations, policy rules and machine-learning models into a visual, largely no-code “decision engine” that risk teams can build and change without waiting on engineering. By June 2026 it had raised roughly $184 million across four rounds — capped by a $110 million Series C led by Growth Equity at Goldman Sachs Alternatives — and reports running hundreds of millions of decisions a month for customers including Mercury, Zilch, Kueski, Rakuten Bank and Allianz across two dozen markets. It is a genuine scale-up (~205-221 staff, undisclosed revenue), now recasting itself from “decision engine” to “agentic decision platform” as the category’s positioning shifts under it.

Founding story

Taktile did not begin as a product idea; it began as a pattern of frustration. Maik Taro Wehmeyer and Maximilian Eber met at Harvard and both landed at QuantCo, an enterprise-AI firm building data-science applications for large companies, a chunk of them in financial services. There, per Wehmeyer’s account to TechCrunch (Nov 2022), the two kept rebuilding the same thing for different clients — the logic that decides who gets approved, at what price, and who gets declined — and realized the reusable asset was the platform underneath, not the bespoke consulting output. In 2020 they left to build it, with the explicit goal of turning the editing of automated decision logic into a self-service act for the risk experts who own the outcomes, rather than a ticket queued behind a data-engineering team.

The founder-market fit is real but narrow: two technical founders who lived the exact pain, deep on data science and regulated-finance decisioning, less obviously seasoned as enterprise operators (a theme the employee reviews return to). Index Ventures backed the seed in 2021 before there was much to see, and has followed every round since — the through-line investor in the story.

How it works

Picture a loan or account application hitting a fintech’s systems. Instead of that request running through hand-coded logic buried in the core banking stack, it is routed to a Taktile “decision flow.” The flow is a graph of nodes the risk team assembles visually: pull a credit-bureau report here, run a KYC/KYB or bank-transaction check there, call a fraud model, apply a policy rule (“decline if debt-to-income > X”), then branch to approve, decline, or refer to a human. Taktile advertises 200+ prebuilt data integrations (bureaus, identity, open-banking, fraud signals) so a team can wire in a new data source without a custom engineering project, and it can host and run the customer’s own machine-learning models inside the flow alongside the rules.

Three capabilities are the actual moat, such as it is. First, backtesting and shadow testing: before a rule change goes live, a team can replay it against historical decisions or run it in parallel with live traffic to see what would have changed — the difference between guessing and knowing. Second, monitoring: live dashboards on approval rates, decline reasons and model performance, with A/B testing to tune cutoffs. Third, governance: version history and audit trails of who changed which rule when — the paperwork a regulator demands. On top of this, Taktile has layered an “AI Copilot” that drafts decision logic from plain-language prompts and, in 2026, “AI nodes” and an “AI Agent Manager” that let flows call large language models for tasks like reading unstructured documents — the mechanism behind its repositioning as an “agentic decision platform.”

Product and business overview

The core product is the Decision Engine — flow builder, integration marketplace, model hosting, backtesting, monitoring. Around it Taktile sells use-case framings rather than separate SKUs: credit underwriting and originations, fraud and transaction monitoring, KYC/KYB onboarding, and — the reason it sits in the insurance slot — insurance underwriting and claims. On the insurance side it pitches underwriting teams a way to combine their policies, pricing models and data on one low-code platform, using OCR/NLP to check submissions for missing disclosures, validate documents, and auto-classify risk so an underwriter can price, approve or escalate faster. Allianz (via Allianz Partners) is the marquee insurance reference, framed around deploying LLMs safely inside mission-critical workflows.

Honest framing: this is a cross-vertical financial-services decisioning platform, not an insurance specialist. Insurance underwriting is one adjacency it has extended into off a credit-and-fintech core, and most of its named traction (Mercury, Zilch, Kueski, Rakuten Bank) is lending and fintech, not carriers.

Business model and pricing

Taktile is enterprise SaaS with no public price list — pricing is demo-and-quote, tailored to decision volume, feature access (AI capabilities cost more) and integration/support needs. Third-party category analysis (Floowed, 2026) buckets Taktile with modern peers running “$50K+/year on demo-only pricing,” versus incumbents (FICO, Provenir, Experian, CRIF) on six-figure annual contracts plus professional services. The economic logic is usage-linked: a platform fee plus scaling with the number of decisions run, which aligns Taktile’s revenue with a customer’s growth — attractive when a fintech is scaling, a liability if the customer’s volumes stall or if a large buyer negotiates a volume floor.

Reported financials are thin and all relative: ARR grew more than 3.5x in 2024 and the customer base roughly quadrupled that year (company, Feb 2025), with no absolute revenue figure disclosed at the Series C. “Hundreds of millions of decisions per month” is the operational metric it leads with instead. For a company that has now raised ~$184M, the absence of a disclosed ARR number is itself a data point.

Traction over time

The shape is a fast-scaling but still-private company: compounding customer and ARR growth off a modest base, geographic expansion into the US and Latin America (a São Paulo office opened to serve LatAm), and a headcount that roughly doubled from the Series B era to 2026. The named logos — Mercury and Zilch in fintech, Kueski in Mexican lending, Rakuten Bank, Monzo, and Allianz in insurance — are credible, but the company has not disclosed how much revenue any of them represent, and “decisions per month” is a usage metric, not a proof of pricing power.

Market analysis

The addressable market depends on which lens you use, and Taktile straddles several. The broad decision-management software market was pegged at roughly $8.05B in 2026, growing to ~$24.3B by 2034 at a ~14.8% CAGR (Fortune Business Insights, 2025). The narrower credit-decisioning-platform segment is smaller but faster: ~$2.84B in the US alone, forecast toward ~$53.9B globally by 2035 at ~21% CAGR (market.us, 2025). On the insurance flank, the AI-underwriting-platform market is estimated at ~$3.55B in 2026 rising to ~$7.12B by 2034 (~7.5% CAGR), inside a much larger “AI in insurance” market of ~$13.45B in 2026 (Fortune Business Insights / IntelMarket, 2026). The structural tailwind is consistent across all of them: financial institutions want faster, more consistent, better-governed automated decisions, and cloud-native platforms are taking share from on-prem incumbents. The structural risk is that “decisioning” is a layer many large buyers can and do build themselves.

Competitive intel

The field splits three ways. Modern platforms — Provenir, GDS Link, and Taktile itself — sell agile, cloud-native decision engines; Provenir is the most direct and best-resourced rival, wider on data marketplace and multi-product coverage, and it competes hard on reference customers while Taktile leans on UX and the agentic-AI narrative. Incumbents — FICO Platform, Experian PowerCurve, CRIF Strategy One — own the installed base inside banks and bureaus; they are slower and pricier (six-figure contracts, 6+ month deployments versus Taktile’s 4-8 weeks), but they carry regulatory trust and inertia that no startup can shortcut. Scoring vendors — Zest AI and Scienaptic — attack from the model side, strong in US credit unions, bundling their own scorecards where Taktile stays model-agnostic. On insurance, Taktile brushes against underwriting-workbench tooling such as Appian’s Connected Underwriting, positioning itself as the decision layer under the underwriter’s screen.

And then the competitor that never shows up on a comparison grid: in-house build. The sophisticated end of Taktile’s market employs risk engineers who can wire bureaus, KYC feeds and models into their own flow, and who may regard a proprietary decision engine as core IP rather than rentable plumbing. Taktile’s whole thesis is that this plumbing is commodity work — governance, integrations, backtesting — that a regulated firm should buy. Whether that holds as foundation-model coding agents lower the cost of building is the open question.

History and evolution

• 2020 — Wehmeyer and Eber leave QuantCo and found Taktile in Berlin to make automated decision logic self-service for risk teams.
• Aug 2021 — $4.7M seed led by Index Ventures; Y Combinator, firstminute Capital, Plug and Play and eight unicorn founders (UiPath, Datadog, GitHub) join.
• Nov 2022 — $20M Series A co-led by Index Ventures and Tiger Global; US expansion accelerates.
• 2024 — Customer base quadruples and ARR grows >3.5x; decision volume reaches hundreds of millions per month; Allianz and Rakuten Bank named as customers.
• Feb 2025 — $54M Series B led by Balderton Capital (Larry Summers among backers), lifting total to $79M; positioning shifts toward “letting risk experts control AI adoption.”
• 2025-2026 — Recasts the product as the “Agentic Decision Platform”; adds AI Copilot, AI nodes and an Agent Manager; opens São Paulo for LatAm; headcount passes 200 across four continents.
• Jun 2026 — $110M Series C led by Growth Equity at Goldman Sachs Alternatives; total raised ~$184M; valuation undisclosed. A Wall Street incumbent leading the round doubles as a distribution signal.

What people say

The case for. Customers and analysts credit Taktile with the sharpest agentic-AI story in the category for scaled fintechs with a risk-engineering team (Floowed, 2026), plus fast deployment and the ability for non-engineers to change decision logic without a release cycle. Employee sentiment is unusually strong: Glassdoor shows ~4.9/5 across 21 reviews with 99% recommending the company (Glassdoor, 2026), praising smart, low-ego colleagues, a product that solves a real problem, strong compensation and a remote-first setup. Marquee logos (Mercury, Zilch, Allianz) and Goldman leading the Series C add external validation.

The complaints. The small review base cuts both ways — 21 Glassdoor reviews is thin, and the negative ones are pointed: allegations of favoritism, feedback used as a control mechanism, “rampant” micromanagement, and leadership that lacks management experience, with one review claiming 2023 saw as many people fired as hired. On product, Gartner Peer Insights and user notes flag that the AI-powered features feel expensive, especially when experimenting or scaling. The structural critique is the loudest, though it rarely comes from a review site: decisioning platforms compete directly with in-house build, and build-vs-buy guidance increasingly argues that a proprietary decision engine and scorecard is exactly the differentiator a lender should own rather than rent — leaving vendors like Taktile to sell the commodity plumbing and the governance overhead that buyers underestimate anyway.

Outlook: the open question

Taktile has built a credible, fast-growing business in a large and structurally growing market, with a clear product, real customers and now a Goldman-backed balance sheet. The question is not whether it can execute — the traction says it can — but whether the category it owns stays a category. Taktile wins if the decision layer proves to be durable, differentiated infrastructure — if the integration maintenance, backtesting rigor, model-risk governance and audit trail are genuinely hard enough, and valued enough, that even sophisticated banks and carriers keep buying rather than building, and if usage-based pricing lets revenue compound as customers scale decision volume. It stalls if foundation-model agents commoditize the build side — if a risk team can cheaply have an AI assemble its own bureaus, KYC feeds and scorecards into a governed in-house flow, collapsing Taktile’s value to the unglamorous parts (compliance paperwork, connector upkeep) that regulated buyers want handled but will not pay a premium for. The insurance angle sharpens rather than resolves this: carriers like Allianz are precisely the buyers with the scale and engineering depth to build, and Taktile’s Allianz relationship is a partnership, not a moat. What to watch: whether it discloses absolute ARR and net retention (the tell on pricing power), and whether the next cohort of logos are carriers and banks — validating the durable-infrastructure thesis — or more fast-scaling fintechs, the customers most able to churn to their own build.

How a challenger would attack it

Weaponize the in-house-build threat Taktile is fighting. Taktile’s own file names its sharpest competitor as “we can build the flow ourselves,” and a challenger would productize exactly that: an open-core, agent-native decisioning framework where a foundation-model agent assembles the customer’s bureaus, KYC feeds and scorecards into a governed flow — selling only the parts regulated buyers genuinely won’t own (audit trail, connector upkeep, model-risk paperwork) at a price that undercuts Taktile’s demo-only, $50K-plus, usage-scaled contracts. The pricing model is the exposed flank: Taktile’s revenue scales with decision volume, and Gartner Peer Insights users already flag the AI features as expensive when experimenting or scaling — a flat-fee or open-source entry converts every volume-sensitive fintech CFO into a lead. Second vector: Taktile’s 200-plus integrations and backtesting engine were built for a pre-LLM world where wiring a bureau was a project; agents collapse that integration cost toward zero, eroding the marketplace’s value faster than the governance layer’s. Third: hit the reference gap. Provenir out-logos Taktile and Taktile discloses no ARR; a challenger that publishes pricing and named carrier deployments attacks a company whose proof points are usage metrics, not revenue.

Same playbook, new buyer

Taktile’s traction is lending and fintech — Mercury, Zilch, Kueski, Rakuten Bank — with insurance as an adjacency and Allianz as a single marquee reference. The most promising shift is to make the carrier the first-class buyer: a decisioning platform built for underwriting workbenches, ACORD submissions, MGA binding-authority rules and claims triage, rather than credit flows with an insurance framing bolted on. Taktile won’t follow easily because its integration marketplace, backtesting semantics and go-to-market muscle are bureau-and-KYC shaped, and its fastest revenue keeps coming from fintechs. Second shift: down-market. Taktile’s quote-only enterprise motion leaves community banks, credit unions below the Zest/Scienaptic scorecard tier, and small MGAs with nothing between Excel and a six-figure contract — a self-serve, published-pricing decision engine for the sub-$50K buyer is a segment the enterprise cost structure can’t chase. Third: regulated verticals beyond finance — healthcare prior-authorization and tenant screening run the same rules-plus-models-plus-audit pattern, and Taktile’s brand, references and data partners are all financial-services-locked.

Sources and further reading

• Taktile Secures $110M in Goldman Sachs-led Series C (BusinessWire, June 24 2026)
• Exclusive: Taktile raises $110 million from Goldman Sachs, Tiger Global (Fortune, June 24 2026)
• Taktile raises $54M to enable risk experts to take control of AI adoption (Taktile / Index Ventures, Feb 27 2025)
• Taktile raises $20M to help fintechs test and deploy decision-making models (TechCrunch, Nov 22 2022)
• Taktile raises $4.7M led by Index Ventures (Index Ventures, 2021)
• The Agentic Decision Platform / Insurance Underwriting solution (Taktile, accessed July 2026)
• Credit Decision Engine Comparison 2026 (Floowed, 2026)
• Taktile Reviews — Pros & Cons of Working at Taktile (Glassdoor, accessed 2026)
• Decision Management Market Size, Share, Growth Report 2034 (Fortune Business Insights, 2025)
• Credit Decisioning Platform Market Size, CAGR of 21% (Market.us, 2025)