Bain rebuilds the agent stack.

What Sage actually is

Bain has operated an internal AI platform under various names since 2021. The early versions were, by the firm's own internal assessment, sophisticated search tools — natural-language interfaces into Bain's proprietary research library, its client engagement archive, and curated third-party data feeds. The system accelerated information retrieval. It did not change what consultants did with information once retrieved. Sage is architecturally different in a way that matters.

Under Valentina Osei, Bain's Chief Technology Officer, the firm rebuilt its platform core around an agent orchestration layer in 2023. Sage accepts structured tasks — a PE due diligence scope, a market entry sizing, a competitive landscape brief — and decomposes them into parallel sub-workflows routed to specialised sub-agents. Each sub-agent draws on a distinct data corpus: Bain's proprietary engagement archive, licensed financial data from PitchBook and Bloomberg, regulatory filings across 22 jurisdictions, and the firm's internally maintained sector databases covering manufacturing, technology, healthcare, and consumer. The sub-agents run in parallel, not in sequence. A due diligence scope that historically took a four-person team eight to twelve days to frame now returns a structured first-draft brief in thirty-six to forty-eight hours. The team spends its time stress-testing the output, not generating it.

The infrastructure partnership underpinning Sage is OpenAI. Bain signed a multi-year commercial agreement with OpenAI in Q3 2023 — one of the largest enterprise AI contracts OpenAI has executed with a professional services firm. The agreement covers dedicated model capacity, fine-tuning access against Bain's proprietary datasets, and a co-development arrangement for specialised reasoning models optimised for financial analysis tasks. Bain's fine-tuned models run on OpenAI's enterprise infrastructure with no external data routing; client information remains inside Bain's environment, which was a non-negotiable condition of the contract given the confidentiality requirements of PE due diligence work.

The private equity agents in production

Private equity due diligence is Bain's most distinctive practice by revenue concentration and by reputational density. The firm runs more PE due diligence mandates than any other global strategy consultancy, and its PE practice accounts for an estimated 35 to 40 per cent of total firm revenue — a concentration that is unusual at the Big Three level and that McKinsey and BCG have historically tried to dilute through broader corporate client portfolios. Bain has not tried to dilute it. It has, instead, built the agent infrastructure to make it defensible at scale.

Sage runs three distinct agent modules inside PE mandates. The first, designated Sage-DD, handles commercial due diligence: market sizing, competitive dynamics, customer concentration analysis, and revenue quality assessment. On a standard commercial DD mandate — a software business being evaluated by a North American buyout fund — Sage-DD ingests the company's management accounts, cross-references market data across Bain's sector databases, benchmarks customer cohort metrics against Bain's proprietary SaaS benchmarking library, and produces a structured commercial DD report with a risk-weighted revenue quality scorecard. The first full production deployment ran in February 2024 on a mid-market healthcare technology company under evaluation by Magnolia Capital Partners, a growth equity fund that has been a Bain PE client since 2018. Sage-DD returned a preliminary commercial DD report in forty-two hours. The engagement team validated, stress-tested, and finalised the output over the following four days. Total elapsed time from scope kick-off to deliverable: six days. The historical benchmark for a comparable scope was fourteen to eighteen days.

The second module, Sage-Fin, handles financial due diligence: quality of earnings analysis, working capital normalisation, covenant mapping across existing debt facilities, and management accounts reconciliation. Sage-Fin integrates directly with the client's data room via a secure API connection, pulls structured financial data without human intermediary extraction, and applies Bain's QoE methodology — built over twenty-two years of PE financial work — as an automated analytical layer. The output is a draft quality of earnings report that a Bain financial DD specialist then reviews, amends, and signs off on. James Ashworth, a Principal in Bain's London PE practice, has described Sage-Fin internally as eliminating "the first week of every financial DD engagement" — the week that previously consisted of data extraction, model building, and preliminary normalisation before any real analysis began.

The third module is the most consequential. Sage-ESG handles environmental, social, and governance due diligence — a workflow that has become structurally mandatory for European PE funds under SFDR and is growing in scope requirements for North American funds ahead of SEC climate disclosure rules. ESG DD was historically outsourced by PE firms to specialist advisors because the cross-jurisdictional regulatory mapping was too complex for generalist consultants to execute efficiently. Sage-ESG changes the economics. The agent monitors regulatory frameworks across 31 jurisdictions in real time, maintains a continuously updated database of SFDR, CSRD, and SEC climate disclosure requirements, and runs automated gap analysis between a target company's disclosed ESG data and the applicable regulatory baseline. For PE funds with SFDR Article 9 mandates — the strictest sustainability classification under the EU framework — Sage-ESG has, in pilot deployments, reduced ESG DD elapsed time from six weeks to nine days.

The OpenAI partnership and what it buys

The Bain–OpenAI commercial agreement is the most significant external partnership in the firm's history and, by the terms of its co-development provisions, structurally different from the enterprise licensing agreements that OpenAI has executed with major corporations. Most large enterprise OpenAI customers receive access to GPT-4 and its successors via the API, with standard fine-tuning options on their proprietary data. Bain's agreement includes a co-development component: OpenAI engineers work directly with Bain's technology team to build reasoning capabilities optimised for the specific analytical patterns of PE due diligence — pattern recognition across financial statement anomalies, precedent-based risk scoring from comparable deal archives, and structured argument generation for investment committee memos.

The practical output of that co-development is a set of Bain-exclusive fine-tuned models that OpenAI maintains on dedicated infrastructure. The models are trained on Bain's engagement archive — anonymised — and the firm's proprietary benchmarking datasets, and they perform materially better on financial analysis tasks than the base GPT-4 models available to standard enterprise customers. Osei's team ran a controlled evaluation in December 2023: the Bain-fine-tuned models outperformed the base model on a set of standardised PE due diligence tasks by 34% on structured output quality, as rated by a panel of Bain senior managers blind to which model produced which output. The benchmark is Bain's own internal standard. It is not an independently audited figure. But within the firm, the evaluation resolved the internal debate about whether the OpenAI co-development investment was justified.

The commercial terms of the agreement have not been publicly disclosed. Internal documents reviewed by INTELAR indicate total committed contract value above $120M over the agreement's initial term, making Bain one of OpenAI's largest enterprise commitments in professional services globally. The agreement also includes provisions for Bain to resell OpenAI capabilities to its PE clients — a licensing layer that allows buyout funds with the scale and sophistication to operate their own AI due diligence infrastructure to access Bain-curated model configurations. Three funds have signed reseller agreements in the first quarter of availability: two large-cap buyout houses based in New York, and one European growth equity firm headquartered in Stockholm. The reseller programme is, in Osei's internal framing, a hedge: if the PE market consolidates toward in-house AI due diligence capability, Bain positions itself as the provider of that capability rather than its casualty.

McKinsey, BCG, and the strategic gap

McKinsey's Lilli rebuild and BCG's internal AI platform represent the most direct competitive comparisons to Sage, and the architectural differences are significant. Lilli is a firm-wide platform — it serves McKinsey's full practice portfolio, from corporate strategy to operations to healthcare. Its agent deployments are broad: practice-area agents in M&A advisory, financial institutions, and technology consulting. Sage is deliberately narrower. Osei's architecture decision — to build depth in PE workflows rather than breadth across practice areas — reflects Bain's revenue concentration and creates a performance advantage that a general-purpose platform cannot easily replicate. A system trained and fine-tuned specifically on PE due diligence analytical patterns produces materially better PE due diligence output than a system trained across all consulting workflows equally. Bain is betting that specialisation compounds.

BCG's position is the most uncertain of the Big Three. The firm has deployed internal AI tooling across its practice areas and, in January 2024, announced a strategic partnership with Google Cloud that covers AI infrastructure and co-development for client-facing products. BCG's partnership with Google maps onto a similar logic to the Bain–OpenAI agreement — dedicated infrastructure, model customisation, co-development provisions — but BCG's comparable PE practice is smaller than Bain's by revenue concentration, and the Google Cloud partnership is oriented more toward enterprise transformation mandates than toward the specialist analytical workflows of PE due diligence. BCG is building a broader AI platform for a broader client set. Whether that positions it to take share from Bain in PE, or merely to defend its corporate advisory practice from McKinsey, is not yet legible from the outside.

The structural competitive risk for McKinsey is that Bain's specialisation advantage in PE due diligence — a market segment where Bain already holds the dominant market share position — may be widening rather than narrowing as Sage reaches production scale. PE due diligence mandates are won on speed, consistency, and the credibility of the firm's sector benchmarking data. Sage improves all three. McKinsey's Lilli, applied to a PE mandate, will produce better output than McKinsey without Lilli, but it will produce that output on a general-purpose inference architecture that cannot match a purpose-built PE agent on the analytical tasks that matter most to a buyout fund CIO making a $500M investment decision. The gap is not permanent. But it is real today, and Bain is operating it at scale while its competitors are still evaluating production architectures.

PE client deployments and the fee model question

Four PE client deployments in Q1 2024 are being watched inside Bain as the commercial test cases that will determine the firm's pricing strategy for agent-primary mandates. The Magnolia Capital Partners healthcare technology DD — the first full Sage-DD production run — completed at 38% of the fee that a comparable traditional mandate would have generated, reflecting the reduced elapsed time and junior consultant hours. Bain's margin on the engagement was 27 percentage points above the firm's PE practice average, driven almost entirely by the reduction in analyst and associate hours billed at below-partner rates. The client paid less. The firm made more per dollar of revenue. The internal accounting team is working through whether that trade is sustainable across the portfolio or a pilot-stage anomaly.

The second deployment — Sage-Fin applied to a carve-out financial DD for a large-cap European industrials transaction — generated a more conventional fee outcome. The client, a Munich-based buyout fund managing €12B in assets, requested that Bain maintain traditional elapsed-time commitments alongside the Sage tooling: the fund's investment committee timeline required the same fourteen-day window as a traditional engagement, and the fund used the Sage efficiency gains to expand scope rather than compress cost. Bain ran a broader financial DD in the same elapsed time for a fee 22% above the original scope estimate, generating an outcome that looks, on the revenue line, identical to a traditional engagement. The internal narrative from the Munich deployment is that Sage creates optionality — clients can use the efficiency gain to get the same scope faster, or the same time to get a broader scope. Both outcomes justify the platform investment.

The fee model debate inside Bain mirrors the one running at McKinsey, and the arguments take the same form. A cohort of senior partners in the North America PE practice, led informally by Thomas Eklund, a Senior Partner in New York who has run more than sixty PE mandates in his eighteen years at the firm, argues that Bain should use the Sage efficiency gain to expand the number of mandates it accepts per quarter rather than to reduce fees on existing mandates. Eklund's thesis: the bottleneck in PE due diligence is not price, it is availability — PE funds go to McKinsey or BCG when Bain is capacity-constrained. If Sage allows each Bain team to run two mandates in the time it previously took to run one, the revenue gain comes from volume, not margin per engagement, and the firm avoids setting a fee floor that the market will exploit. The argument has not yet won the firm-wide debate, but it has the most traction among senior partners closest to the PE client relationship.

What to watch

Five signals that will determine whether Bain's Sage platform becomes a durable structural advantage or stalls on the pricing and partnership tensions it has already surfaced.

• The outcome of Bain's Q2 2024 partner compensation round. PE practice margins on Sage-primary engagements are running 22 to 27 percentage points above legacy averages. If those margins flow through to partner draws at a rate that visibly exceeds compensation in practices that have not yet deployed Sage, the platform adoption inside the firm accelerates on economic incentive alone. If the compensation committee smooths the differential across the partnership — which it has done in prior years when single-practice performance diverged — the incentive weakens and Sage adoption decelerates outside the PE practice.
• Sage-ESG's first Article 9 fund certification deployment. The SFDR Article 9 classification is the EU's most demanding sustainability standard for PE funds. If Sage-ESG's nine-day ESG DD cycle successfully supports a full Article 9 certification process — rather than a compliance gap analysis — it validates the module's regulatory-grade output quality and creates a distinct commercial offering that Bain's European PE competitors cannot match on speed.
• Whether McKinsey discloses a PE-specific agent rebuild. McKinsey's Lilli platform has deployed in CIB, FIG, and BTO practices. A disclosure of PE due diligence–specific agent modules — particularly one with comparable QoE and ESG DD capability — signals that the competitive window Bain currently holds is narrowing. McKinsey has the proprietary data and the engineering infrastructure to build PE agents. The question is whether it has the organisational incentive to prioritise a practice where it is not the market share leader.
• The reseller programme's fund count by year-end 2024. Three PE funds signed reseller agreements for Bain-curated OpenAI model configurations in Q1. If the programme reaches twenty funds by December, Bain has established a recurring software revenue stream that is structurally independent of its consulting engagement pipeline — and has begun the transition to the consulting-as-software model without announcing it publicly.
• Eklund's volume thesis versus the fee-reduction camp's pricing experiment. The internal argument between running more mandates at existing fees versus fewer mandates at lower fees will be decided by Q3 data. If the two Sage-primary pilots with compressed fees — including the Magnolia engagement — generate repeat business at the lower fee level, the pricing conservatives' argument collapses. If Magnolia returns for the next mandate expecting the same 38% fee discount and Bain cannot hold the line, Eklund will have been correct.