AI application-layer startup radar: July 29–August 5, 2026

Executive read

The July 29–August 5 window produced 13 public application-layer funding disclosures: eight dollar-denominated rounds totaling $470.4 million, plus five euro-denominated rounds totaling €39.91 million. The totals are sums of current-round disclosures only; currencies are kept separate rather than converted at an arbitrary exchange rate.

The headline dollars were concentrated in three follow-on rounds: Simile’s $200 million Series B, HappyRobot’s $150 million Series C, and Faye’s $50 million Series C. The more useful early-stage signal is elsewhere: June raised $20 million before emerging from stealth to attack the enterprise AI deployment bottleneck; Dili and Intropy are turning difficult compliance and spare-parts data into executable vertical workflows; and Shiplog, Aavalynx, Aflabox, and conmeet are attaching agents or AI decision systems to narrower operating domains.

This issue separates newly surfaced or early-stage companies from mature-company financing signals. A large Series C is a capital-market signal, not evidence that the company itself first appeared this week. The diligence question across both groups is the same: does the product own a workflow, data asset, or operating relationship that a general-purpose model vendor cannot copy by adding a feature?

Quick scan

Company	Sector	Window signal	Product positioning	Founding-team background
June	AI agents / enterprise deployment	$20M pre-seed; Aug. 3	Scans legacy systems and maps the steps needed to deploy agent-powered enterprise workflows	Efrat Rapoport, Ohad Hen, Barak Goldstein, and Idan Tsitiat previously founded Bonobo AI, acquired by Salesforce; the team later worked on Salesforce AI initiatives
Henry AI	Vertical SaaS / commercial real estate	$16.5M Series A; July 29	Automates offering memorandums, underwriting, pitch decks, buyer lists, and other CRE back-office work	Sammy Greenwall (CEO) and Adam Pratt (CTO); the announcement does not disclose prior affiliations
Dili	Vertical SaaS / infrastructure compliance	$15M Series A; July 30	Extracts evidence from project documents and checks construction and clean-energy compliance rules	Anand Chaturvedi is co-founder and CEO; Dili was in Y Combinator’s Summer 2023 batch; fuller prior affiliations were not disclosed in the opened source
Intropy	Vertical SaaS / spare parts	$11M seed; July 30	Executes demand forecasting, inventory, obsolescence, and pricing decisions inside ERP workflows	Franziska Kirschner and YihKai Teh previously worked at Tractable and are named on more than 10 patents; Kirschner is an Oxford-trained physicist, and Teh an AI academic from UCL
Intelligence / Design Arena	Consumer AI / model evaluation	$7.9M seed; Aug. 3	Uses human preference data from visual-output comparisons to improve media-generating models	The source names co-founder Grace Li and says the team began as college friends around graduation in 2025; no fuller founding history is disclosed
Simile	Consumer research / synthetic users	$200M Series B at $2B; July 30	Simulated users for marketing and product research	Founded by Stanford PhD Joon Sung Park, whose dissertation involved the Smallville simulated-agent project; no other founder background is given in the source
HappyRobot	AI agents / enterprise operations	$150M Series C at $1.2B; Aug. 4	Agents automate workflows across voice, email, documents, and the web	Pablo Palafox is identified as co-founder and CEO; the opened source does not provide a fuller founding-team history
Faye	Consumer AI / travel care	$50M Series C; Aug. 5	Travel protection, claims, policy management, and travel-fintech workflows with an autonomous-care roadmap	Elad Schaffer is identified as co-founder and CEO; the release does not name a fuller founding team
Shiplog	Vertical SaaS / B2B customer success	€807.6k pre-seed; Aug. 4	Agent Ada evaluates each customer and chooses next actions across the lifecycle	Khushi Mehta (CEO) and Mehdi Gribaa (CTO); prior affiliations were not disclosed
Aavalynx	Vertical SaaS / legal	€1.75M pre-seed; Aug. 4	Sisu supports dispute assessment, chronology building, document browsing, and metadata extraction	Hanna Roos is founder and CEO; the source does not disclose previous affiliations
Aflabox	Vertical AI / food safety	€1.35M seed; Aug. 3	Combines UV imaging, AI, and geolocated data to detect aflatoxin risk in grain supply chains	Fabrizio Cardillo and Luca Alinovi; prior affiliations were not disclosed
conmeet	Vertical SaaS / construction	€6M seed; Aug. 5	Operating system for project management, procurement, scheduling, site operations, finance, and invoicing	Benedikt Kisner, Leandro Ananias, and Lennart Eckerlein; prior affiliations were not disclosed
Moss	Vertical SaaS / finance	€30M Series C; Aug. 5	Spend management platform expanding into AI agents for finance teams	Founded in 2019 by Ante Spittler, Anton Rummel, Ferdinand Meyer, and Stephan Haslebacher

Traction figures below are company- or founder-reported unless explicitly identified otherwise. A missing background field means the opened source did not provide it; it is not an inferred biography.

Funding profiles

June: the application layer starts with deployment, not the model

June emerged from stealth with a $20 million pre-seed led by Marc Benioff’s Time Ventures, with additional backing from Michael Dell, Aaron Levie, and George Kurtz. The company declined to disclose a valuation. Its product scans a company’s existing systems, finds bottlenecks, maps the steps required to deploy agents, and can then build the workflow task by task.

The founding-team signal is unusually clear: Rapoport, Hen, Goldstein, and Tsitiat previously built Bonobo AI, which Salesforce acquired; they then spent several years working on Salesforce’s AI initiatives. June’s wedge is therefore not another general-purpose agent interface. It is an attempt to productize the messy integration and process-mapping work that enterprise AI adoption currently pushes onto consultants and forward-deployed engineers. The diligence question is whether June can make that work repeatable without becoming a services business in disguise. 1

Henry AI: a narrow workflow with measurable production time

Henry AI announced a $16.5 million Series A led by FirstMark Capital, with Thomson Reuters Ventures, Y Combinator, Susa Ventures, 1Sharpe, StoryHouse Ventures, Pioneer Fund, RXR Arden Digital Ventures, Karman Ventures, and Coalition Operators participating. The company also launched Henry Deal, which turns a firm’s institutional memory into transaction deliverables for commercial real estate.

The company says more than 150 firms use the platform, including teams from all the largest commercial real estate brokerages. It reports more than 20,000 client-ready deliverables tied to more than $150 billion in underlying deal value; analyst production time is down 90%, and work that once took 15 hours now takes roughly 30 minutes of human review. Those are strong workflow-adoption signals, but they are self-reported. The key question is whether Henry owns the firm’s transaction context deeply enough to defend against a brokerage’s internal build or a horizontal document agent. 2

Dili: deterministic rules around an LLM extraction layer

Dili raised $15 million in Series A funding led by Khosla Ventures, with Allianz, Rebel Fund, Darren Bechtel of Brick & Mortar Ventures, and Y Combinator’s Garry Tan participating. The round followed a $6.7 million seed, taking total capital raised to $21.7 million. Dili targets the overlapping rules attached to construction and infrastructure projects, including prevailing-wage, apprenticeship, OSHA, and EPA requirements.

The architecture is the important part for diligence: contemporary models extract and structure data from unstructured documents, then a deterministic system applies the relatively static compliance rules. The company says the system is used across about 700 projects, with roughly half sold as software and half delivered through an outsourced compliance-service model. Anand Chaturvedi, co-founder and CEO, is the named team signal; Dili was a Y Combinator Summer 2023 company. The product’s defensibility will depend on rule coverage, auditability, and how quickly the services mix can become software revenue. 3

Intropy: from spare-parts data to decisions inside ERP

Intropy raised an $11 million seed round led by Felix Capital, with Quiet Capital, General Catalyst, and firstminute capital participating. It is building an AI-native operating system for the spare-parts industry: demand forecasting, inventory distribution, obsolescence management, and dynamic pricing are integrated into existing ERP and operational systems so the software can execute decisions rather than only display recommendations.

The founders, Franziska Kirschner and YihKai Teh, previously worked together at Tractable and are credited with more than 10 patents applying AI to spare parts. The company describes Kirschner as an Oxford-trained physicist whose research appeared in Nature, and Teh as a University College London AI academic from Malaysia. Intropy says its technology has processed more than $10 billion in parts demand and that customers have seen more than 10x ROI; both are company-reported. The diligence question is whether its domain data and ERP execution create a compounding advantage, or merely a difficult implementation that larger supply-chain suites can absorb. 4 5

Intelligence / Design Arena: human taste becomes a model input

Intelligence, the company behind Design Arena, raised a $7.9 million seed led by Index Ventures, with Conviction, A*, Valkyrie, and others participating. Design Arena presents users with A-versus-B choices across websites, images, and other visual formats; the resulting preference data is sold as feedback for media-generating AI models.

Co-founder Grace Li says the company began with college friends around graduation in 2025 after they found that AI-generated games could be functional without being fun. Intelligence says Design Arena has 5.3 million users and is generating $60 million in ARR, but the source provides no independent verification of either number. The useful signal is not the router-like consumer interface by itself; it is the possibility of a durable, logged preference dataset that automated benchmarks cannot easily reproduce. The counter-signal is Yupp’s shutdown after raising $33 million, showing that human-evaluation usage does not automatically become a durable business. 6

Simile: a $200M bet on simulated market research

Simile announced a $200 million Series B at a $2 billion valuation, only five months after emerging from stealth with a $100 million Series A led by Index Ventures. The new round was led by Greenoaks, with Index, Hanabi, Bain Capital Ventures, A*, Factory, Definition, and CVS Health Ventures participating; CVS is also described as a marquee customer.

The product offers simulated users for marketing and product research. Founder and CEO Joon Sung Park holds a Stanford PhD, and his dissertation included Smallville, a project in which AI agents lived simulated human lives. This is a clean example of a company selling a behavioral proxy rather than a workflow agent. The diligence burden is correspondingly high: the value of simulated responses must be tested against real-world holdouts, not inferred from the valuation or the ambition to simulate all eight billion people. 7

HappyRobot: agents move across the operating surface

HappyRobot raised $150 million in Series C at a $1.2 billion post-money valuation, led by Prysm Capital and co-led by Eurazeo. The round included a16z, Base10, Y Combinator, Koch Disruptive Technologies, KFund, Orange, T.Capital, Bankinter, Endeavor Catalyst, and Wave-X. The company says it has raised around $200 million in total.

Its agentic platform automates enterprise-operations and supply-chain workflows across voice, email, documents, and the web. HappyRobot reports more than 150 enterprise customers, fivefold growth since its Series B late last year, and one customer automating 28,000 hours of work per month. It also reports 70%+ autonomous resolution and a typical 4–12 week deployment. The signal is breadth of operating surface; the risk is that a horizontal agent platform becomes a collection of integrations whose economics are hard to defend. Pablo Palafox is the named co-founder and CEO in the opened source. 8

Faye: consumer AI attached to a regulated service relationship

Faye raised $50 million in Series C, led by Madrona, with BRM, Portage, F2 Venture Capital, Viola Ventures, and Lumir Ventures participating. The company says total funding has reached $100 million. Its product combines travel protection, claims and policy management, distribution for travel agencies, online travel agencies and airlines, and the Faye Wallet travel-fintech product.

Co-founder and CEO Elad Schaffer says the company expects AI to resolve more than half of claims autonomously by year-end, with 75% of remaining claims completed on first touch. Faye also reports doubled revenue over the past year and more than 20 industry awards. These are company-reported traction signals. The application-layer thesis is stronger than “AI for travel” alone: Faye owns the policyholder relationship and the claims workflow, giving it a regulated data and distribution surface that a generic consumer assistant does not have. 9

Shiplog: an early agent with a measurable pilot surface

Shiplog raised more than €807.6k in pre-seed funding from Kima Ventures, Project Europe, Purple, No Label Ventures, 100IN, and Station F Fund. Its AI agent, Ada, evaluates each B2B SaaS customer in real time, chooses the next best action, and personalizes the customer journey across tools including Salesforce, HubSpot, Snowflake, Shopify, and Stripe.

Co-founders Khushi Mehta and Mehdi Gribaa are building from Paris. The company says an early pilot analyzed more than 4 million events across roughly 1,000 customers; a separate account describes the company as rolling out partner pilots and expanding integrations. That is an early signal, not proof of retention or expansion revenue. The diligence question is whether Ada changes customer outcomes or remains an analytics layer with an agent label. 10

Aavalynx: legal AI built around the dispute timeline

Aavalynx raised €1.75 million in pre-seed funding from Omega Ventures, Two Ravens, senior law-firm partners, and a former Amazon European head. Its Sisu platform supports dispute resolution from early assessment through hearings, including chronology construction, document browsing, and metadata extraction.

Founder and CEO Hanna Roos is the only named founder in the opened announcement. Aavalynx reports early data showing roughly 30x ROI in saved damages, legal fees, and interest, or up to 200x when rescued commercial opportunities are included. These figures are company-reported and should be tested against case-level outcomes; the product’s more durable signal is its position in a high-value, document-heavy workflow. 11

Aflabox: a vertical AI application with a hardware edge

Aflabox raised €1.35 million in seed funding from FoodSeed and Farming Future. Its portable platform combines UV imaging, AI, and geolocated data to detect mycotoxin and aflatoxin risk, assess grain quality, and map supply-chain risk.

Fabrizio Cardillo and Luca Alinovi are the named founders. The company says it was founded in July 2024, has supplied 20 devices to the World Food Programme in Kenya, completed an initial validation phase, and filed a patent application. The hardware component makes this a boundary case for an application-layer radar; it is included because the product’s customer value is the decision and risk layer built on top of field data, not a general-purpose sensing or model platform. 12

conmeet: an operating system for trades, not a generic construction copilot

conmeet raised an oversubscribed €6 million seed round from Reimann Investors Venture Capital and Smedvig Ventures, following pre-seed backing from May Ventures. Its platform combines project management, procurement, scheduling, site operations, documentation, finance, and invoicing for trades and construction businesses with 10 to 500 employees.

The founders are Benedikt Kisner, Leandro Ananias, and Lennart Eckerlein. The company says it was founded in 2023 and is expanding across the DACH region. The application-layer signal is workflow breadth inside a defined operator: the risk is implementation complexity and competition from incumbent construction-management software. 13

Moss: the finance-suite incumbent adds an AI layer

Moss closed a €30 million Series C led by Portage, with existing investor Cherry Ventures participating. It is a Berlin-based spend-management platform covering corporate cards, invoices, reimbursements, budgets, approvals, and accounting, while expanding into a Finance AI suite and finance agents.

Moss was founded in 2019 by Ante Spittler, Anton Rummel, Ferdinand Meyer, and Stephan Haslebacher. This is a mature-company financing signal rather than a newly emerged startup: its value for this radar is the evidence that investors are still funding AI expansion inside a live finance-workflow product. 14

Cluster notes for diligence

1. The agent is becoming an implementation surface

June, HappyRobot, Shiplog, and Henry AI all put the product between a model and a business system. They do not compete only on model quality. They compete on process mapping, permissions, data normalization, deployment time, and the cost of getting a customer to a reliable production state. That makes implementation evidence more important than a demo: June points to enterprise deployment pain, HappyRobot reports 4–12 week go-lives, Shiplog points to a quantified pilot surface, and Henry points to completed deliverables.

The open question is margin. If each deployment requires bespoke data cleanup, the company may be selling professional services with an AI interface. If the product can turn the first implementation into reusable schemas, playbooks, and evaluation data, the same work can become a compounding software asset.

2. Vertical software is moving from prediction to action

Dili’s deterministic compliance engine, Intropy’s ERP execution, Aavalynx’s dispute workflow, Aflabox’s field-risk decisions, conmeet’s construction operating system, and Moss’s finance suite all attach AI to a decision that already has an owner and a cost of delay. This is a stronger diligence frame than simply counting “AI features.”

The recurring questions are: who owns the source-of-truth data; who is authorized to act on the recommendation; how is an error audited; and does the product replace a manual handoff or merely add another dashboard? The answers determine whether the AI layer captures budget or remains an experimental add-on.

3. Human preference and synthetic behavior are two different bets

Intelligence sells logged human preferences to model builders; Simile sells simulated responses for product and market research. Both monetize a proxy for human judgment, but the evidence burden differs. Design Arena can test its data against downstream model-improvement outcomes. Simile must show that simulated users predict real behavior better or faster than conventional panels. Neither company’s funding size is a substitute for that validation.

Product-launch checkpoints, not funding events

Product Hunt’s daily leaderboards surfaced several application-facing launches during the same period. These are positioning signals only: they do not establish funding, founder history, customer traction, or a company’s long-term survival.

NudgeForMe — Aug. 1: an AI follow-up agent for missed email opportunities. 15
Zinley — Aug. 2: a personal AI representative for calls, email, and tasks. 16
Airtop for Google Ads Automation — Aug. 3: campaign building, spend optimization, and reporting. 17
Hey Noah — Aug. 4: a proactive AI executive assistant for founders. 18
Framer AI Agents — Aug. 5: AI-assisted design and publishing for professional websites. 19

No Product Hunt item above is included in the funding totals. The leaderboards are useful for seeing where application positioning is appearing—follow-up, personal representation, paid acquisition, executive assistance, and site production—but they are weak evidence for diligence until primary product, team, and customer information is available.

Sourcing and scope notes

Window: public disclosures dated July 29 through August 5, 2026, matching the scheduled weekly issue. Where a source supplied only a calendar date rather than an hour, this article makes no hour-level claim.
Counting: the headline total uses the current round in each disclosure. Simile’s prior $100 million Series A, Dili’s prior $6.7 million seed, and Faye’s cumulative $100 million are shown as context, not added to the current-round total.
Currencies: USD and EUR are kept separate; no exchange-rate conversion is used. Shiplog’s reported €807.6k is counted in the euro subtotal.
Classification: this radar excludes model-layer companies and infrastructure providers. Aflabox is a stated boundary case because its application combines software with a portable device; Neuraspace was not included because its disclosure emphasizes proprietary sensing infrastructure and defense capabilities.
Evidence: media reports and company releases are linked at the fact unit. Founder backgrounds are limited to what the opened sources disclosed; no missing affiliation is filled from an unverified database.

The next issue will continue the weekly application-layer scan with the same separation between early signals, mature-company financing, and launch-positioning checkpoints.