Whereas synthetic intelligence has stormed into regulation companies and accounting practices with billion-dollar startups like Harvey main the cost, the worldwide consulting business—a $250 billion behemoth—has remained stubbornly analog. A London-based startup based by former McKinsey consultants is betting $2 million that it could crack open this resistant market, one Excel spreadsheet at a time.
Ascentra Labs introduced Monday that it has closed a $2 million seed spherical led by NAP, a Berlin-based enterprise capital agency previously often known as Cavalry Ventures. The funding comes with participation from notable founder-angels together with Alan Chang, chief govt of Fuse and former chief income officer at Revolut, and Fredrik Hjelm, chief govt of European e-scooter firm Voi.
The funding is modest by the requirements of enterprise AI — a sector that has seen funding rounds routinely attain into the a whole bunch of thousands and thousands. However Ascentra's founders argue that their targeted method to a slender however painful downside may give them an edge in a market the place broad AI options have repeatedly failed to realize traction.
Consultants spend numerous hours on Excel survey evaluation that even high companies haven't automated
Paritosh Devbhandari, Ascentra's co-founder and chief govt, spent years at McKinsey & Firm, together with a stint at QuantumBlack, the agency's AI and superior analytics division. He is aware of intimately the late nights consultants spend wrestling with survey knowledge—the type of quantitative analysis that varieties the spine of personal fairness due diligence.
"Earlier than beginning the corporate, I used to be working at McKinsey, particularly on the personal fairness workforce," Devbhandari defined in an unique interview with VentureBeat. The work, he stated, includes analyzing encoded survey responses from clients, suppliers, and market individuals throughout potential acquisitions.
"Consultants sometimes spend a whole lot of time doing this in Excel," he stated. "One of many issues that shocked me, having labored at a few totally different locations, is that the workflow — even at the very best companies — actually isn't that totally different from among the boutiques. I at all times anticipated there can be some smarter method of doing issues, and sometimes there simply isn't."
That hole between expectation and actuality grew to become the muse for Ascentra. The corporate's platform ingests uncooked survey knowledge information and outputs formatted Excel workbooks full with traceable formulation — the type of deliverable a junior affiliate would spend hours setting up manually.
AI has remodeled authorized work however consulting presents distinctive technical challenges which have blocked adoption
The disparity between AI adoption in regulation versus consulting raises an apparent query: if the consulting market is so giant and the workflows so guide, why hasn't enterprise capital flooded the area the best way it has authorized tech?
Devbhandari provided a frank evaluation. "It's not like individuals haven't tried," he stated. "The highest of the funnel in our area is crowded. Once we converse to our consulting purchasers, the companions say they get one other pitch deck of their LinkedIn inbox or e-mail each week—generally a number of. There are many individuals making an attempt."
The limitations, he argued, are structural. Skilled providers companies transfer slowly on expertise adoption, demanding intensive safety credentials and buyer references earlier than granting even a pilot alternative. "I feel that's the place 90% of startups in skilled providers, writ giant, fall down," he stated.
However consulting presents distinctive technical challenges past the gross sales cycle. Not like authorized work, which largely includes textual content paperwork that fashionable giant language fashions deal with nicely, consulting spans a number of knowledge modalities — PowerPoint shows, Excel spreadsheets, Phrase paperwork — with info that may be tabular, graphical, or textual.
"You’ll be able to have a number of codecs of Excel in itself," Devbhandari famous. "And that's a giant distinction to the authorized area, the place you can have a multi-purpose AI agent, or assortment of brokers, which might truly do a whole lot of the duties that legal professionals do everyday. Consulting is the alternative of that."
Ascentra's personal fairness focus displays a calculated wager on repeatable workflows
Ascentra's technique hinges on excessive specificity. Quite than making an attempt to automate the total spectrum of consulting work, the corporate focuses solely on survey evaluation inside personal fairness due diligence — a distinct segment inside a distinct segment.
The logic is each technical and business. Non-public fairness work tends to be extra standardized than different consulting engagements, with comparable analyses recurring throughout offers. That repeatability makes automation possible. It additionally positions Ascentra towards a much less formidable aggressive set: even the most important consulting companies, Devbhandari claimed, lack devoted inside instruments for this explicit workflow.
"Survey evaluation automation is so particular that even the most important and finest companies haven't developed something in-house for it," he stated.
The corporate claims that three of the world's high 5 consulting companies now use its platform, with early adopters reporting time financial savings of 60 to 80 % on lively due diligence initiatives. However there's a notable caveat: Ascentra can not publicly identify any of those purchasers.
"It's a really personal business, so in the meanwhile, we will't announce any purchasers publicly," Devbhandari acknowledged. "What I can say is that we're working with three of the highest 5 consulting companies. We've handed pilots at a number of organizations and have submitted enterprise instances for enterprise rollouts."
Eliminating AI hallucinations turns into essential when billion-dollar offers cling within the stability
For an AI firm promoting into quantitative workflows, accuracy is existential. Consultants delivering evaluation to non-public fairness purchasers face huge strain to be exact—a single error in a monetary mannequin can undermine credibility and, probably, billion-dollar funding choices.
Devbhandari described this as Ascentra's central design problem. "Consultants require a really, very excessive diploma of constancy after they're doing their evaluation," he stated. "So with quantitative knowledge, even when it's 95% correct, they’ll revert to Excel as a result of they realize it, they belief it, and so they don't need there to be any margin for error."
Ascentra's technical method makes an attempt to deal with this by limiting the place AI fashions function inside the workflow. The corporate makes use of GPT-based fashions from OpenAI to interpret and ingest incoming knowledge, however the precise evaluation depends on deterministic Python scripts that produce constant, verifiable outputs.
"What's totally different is the steps that comply with are deterministic," Devbhandari defined. "There's no room for error. There's no hallucinations, and the Excel author that we've related to the product on the again finish converts this evaluation into Excel components, that are dwell and traceable, so consultants can get that assurance that they will comply with together with the maths."
Whether or not this hybrid method delivers on its promise of eliminating hallucinations whereas sustaining helpful AI capabilities might be examined because the platform scales throughout extra complicated use instances and consumer environments.
Enterprise safety certifications give Ascentra an edge over much less ready opponents
Promoting software program to main consulting companies requires clearing an unusually excessive safety bar. These organizations deal with delicate consumer knowledge throughout industries, and their vendor safety assessments can take months to finish.
Ascentra invested early in acquiring enterprise-grade certifications, a strategic alternative that Devbhandari framed as important desk stakes. The corporate has achieved SOC 2 Sort II and ISO 27001 certifications and claims to be underneath audit for ISO 42001, an rising customary for AI administration techniques.
Information dealing with insurance policies additionally mirror the sensitivity of the goal market. Shopper knowledge is deleted inside 30 to 45 days, relying on contractual phrases, and Ascentra doesn’t use buyer knowledge to coach its fashions.
There's additionally an argument that survey knowledge carries considerably decrease sensitivity than different consulting supplies. "Survey knowledge is exclusive in consulting knowledge as a result of it's collected in the course of the course of a undertaking, and it’s market knowledge," Devbhandari famous. "You interview individuals out there, and also you accumulate a bunch of information in an Excel, versus—you have a look at Rogo or among the different finance AI startups—they use consumer knowledge, so financials, which is confidential and strictly private."
Per-project pricing aligns with how consulting companies truly spend cash
Ascentra's pricing mannequin departs from the subscription-based method that dominates enterprise software program. The corporate expenses on a per-project foundation, a construction Devbhandari stated aligns with how consulting companies allocate budgets.
"Venture budgets are in consulting set on a per undertaking foundation," he defined. "You'll have central budgets that are for issues like Microsoft, proper, very central issues that each workforce will use the entire time. After which you have got undertaking budgets that are for the groups which might be utilizing particular assets, groups or merchandise these days."
This method might ease preliminary adoption by avoiding the necessity for central IT procurement approval, however it additionally introduces income unpredictability. The corporate's success will depend upon changing project-level utilization into broader enterprise relationships—a path Devbhandari recommended is already underway by submitted enterprise instances for enterprise rollouts.
AI might not get rid of consulting jobs, however it should basically rework what consultants do
Maybe essentially the most fascinating stress in Devbhandari's imaginative and prescient considerations what AI finally means for consulting employment. He pushed again on predictions that AI will get rid of consulting jobs whereas concurrently describing an business on the cusp of elementary transformation.
"Individuals love to speak about how AI goes to take away the necessity for consultants, and I disagree," he stated. "Sure, the function will change, however I don't suppose the business goes away. I feel the very best options will come from individuals inside the business constructing merchandise across the work they know."
But he additionally painted an image of dramatic change. "In the meanwhile, you have got a giant consumption of graduates who simply do—for essentially the most half, you understand, they’ve the strategic work as a part of what they do, however additionally they have a whole lot of work in Excel and PowerPoint. I feel in a number of years' time, we'll look again at these instances and suppose, you understand, very, very totally different."
The sincere reply, he acknowledged, is that nobody actually is aware of how this performs out. "I don't suppose even AI leaders actually know what that appears like but," he stated of whether or not productiveness beneficial properties will translate to extra work or fewer staff.
Ascentra plans to make use of seed funding to broaden its U.S. presence and go-to-market workforce
The $2 million will primarily fund Ascentra's growth into the USA, the place greater than 80 % of its clients are already based mostly. Devbhandari plans to relocate there personally as the corporate builds out go-to-market capabilities.
"One of many issues that we've actually seen is that with consulting being an American business, and I feel America being an important place for innovation and making an attempt new issues, we've undoubtedly drawn ourselves to the U.S.," he stated. "American hires are very costly, and I'm positive that a whole lot of the increase will go in the direction of that."
The seed spherical represents a wager by NAP on what its co-founder Stefan Walter known as an overdue disruption. "Whereas most data work has been reshaped by new expertise, consulting has remained stubbornly guide," Walter stated. "AI received't substitute consultants, however consultants utilizing Ascentra would possibly."
The startup now faces the arduous work of changing pilot wins into lasting enterprise contracts
Ascentra enters 2026 with momentum however no assure of success. The corporate should rework pilot applications at elite companies into sticky enterprise contracts — all whereas keeping off the inevitable well-funded opponents who will flood into the area as soon as the chance turns into simple. Its intentionally slender give attention to survey evaluation supplies a defensible beachhead, however increasing into adjoining workflows would require constructing totally new merchandise with out sacrificing the area experience that Devbhandari argues is the corporate's core benefit.
Oliver Thurston, Ascentra's co-founder and chief expertise officer, who beforehand led machine studying at Mathison AI, provided a clear-eyed evaluation of the problem. "Consulting workflows are uniquely complicated and troublesome to construct merchandise round," he stated in an announcement. "It's not shocking the area hasn't modified but. It will change although, and there's little doubt that the business goes to look fully totally different in 5 years' time."
For now, Ascentra is putting a targeted wager: that the consultants who as soon as spent their nights formatting spreadsheets would be the ones who lastly convey AI into an business that has lengthy resisted it. The irony is difficult to overlook. After years of advising Fortune 500 firms on digital transformation, consulting might lastly need to take its personal medication.
