Vendor Strategy

Zip, Fairmarkit, Pactum and Coupa do different jobs. If you own a suite, test its AI first, then buy a specialist only for the gap you measured.

Meta has acquired Conductor. Cloud workspaces go read-only on November 13, deletion starts November 20, and the local Mac app gets updates only until early 2027.

OpenAI will acquire Pierre and move Code Storage's features into Codex. Customers can stay after closing, with no end date, so map your exit now.

Kimi K3.1 Pro's 1.56 TB weights are public and free for internal use. An eight-GPU B300 node breaks even with the $2/$10 API near 13 billion input tokens a month, and legal and tax scores fell from K3.

Docusign announced its acquisition of Suplari with five undated plans and nothing for current customers. Microsoft told those customers "no change" in 2021, then divested the company in 2023.

CrowdStrike plans to move Dreadnode's AI security testing onto Falcon, but neither company has addressed existing contracts, standalone sales or on-prem deployment. Get those terms in writing and keep a second tester for outside evidence.

Marvell's $115 billion all-stock deal for Lumentum leaves customer agreements in place only until close. Buyers have about 12 to 18 months to qualify second-source optics and secure post-close supply terms.

Affle's $420M cash deal for Perion comes with a post-close review of business contracts. Advertisers and agencies on Perion One or Outmax should pin down rates and data terms before the early 2027 close.

Aramco closed on 49% of HUMAIN, and the EU filing describes joint control with PIF. Anyone buying HUMAIN compute, cloud or ALLaM should settle data separation, capacity priority and exit terms before signing.

Certify to ISO 42001 when a customer requires it in writing. One accredited auditor's figures imply $46K to $80K+ in audit fees over three years, and the certificate gives no EU AI Act presumption of conformity.

LLM API SLAs pay credits for failed requests and say little about slow ones. Compare OpenAI, Azure, Bedrock, Vertex and Anthropic, then negotiate definitions and build a fallback that triggers on latency.

MentorcliQ has acquired Paris-based Neobrain, which hosts customer data with OVH in France. No price or customer terms were disclosed, so Neobrain buyers should get hosting, roadmap and integration commitments in writing.

Deepgram is acquiring Daily, the maker of Pipecat. The BSD-2 framework stays open and gets a promised foundation, but Daily's WebRTC, telephony and Pipecat Cloud contracts change owner.

Salesforce signed a deal to acquire Rune Technologies, maker of the TyrOS edge logistics platform, reportedly for $375M. TyrOS users should pin down the deal structure, the fate of Rune's $99M Army IDIQ and where edge data will sync before close.

Agentforce Flex Credits bill $0.10 per action, so a service agent's cost per resolved case runs from $0.50 to $3.33 depending on how many steps it takes. Help Agent's $2 per resolution wins above about 12 actions at 60% resolution, and the $2 per-conversation SKU never wins.

Apple is hiring Ascend.io's team and licensing its IP, and Ascend says it is winding down with no shutdown date. Pipeline code lives in your Git repo but runs only on Ascend's runtime, so migration should start now.

Mistral Large 4's 1.05T-parameter weights are on Hugging Face. At FP8 they fit one eight-GPU DGX B200 node, internal use is free, and serving the model to outside customers needs a separate commercial licence from Mistral.

NVIDIA AI Enterprise lists at $4,500 per GPU per year. For bare-metal inference, vLLM or SGLang on your own stack, or Red Hat AI for a support contract, replaces it; vGPU and production NIM are where you still need it.

GLM-5.3 on Cloudflare Workers AI returned other customers' data in 155 long-context responses because an undersized KV pool tripped an SGLang disaggregation bug. Upgrade SGLang to v0.5.22 and ask your inference provider how KV pages are isolated.

Claude's $1-per-user federal price ends October 31. The new GSA OneGov terms give federal, state, local and tribal buyers 40% off metered tokens through January 2029, so agencies need a spend cap and a cost-per-task comparison against OpenAI's 50% offer.

Since the Digital Omnibus, Article 4 asks for measures and a record, not a set level of knowledge. Build a role map on free vendor academies and pay for specialist modules only where staff oversee AI.

The Bank of England and FCA's 2026 survey finds 80% of UK financial firms run AI agents and 2% grant broad autonomy. Vendor AI now makes up 53% of high-materiality systems, and a quarter of the 48 firms asked let vendor updates reach production automatically.

Zafin signed a deal to buy Naehas, the offer management and disclosure automation platform used by large U.S. banks. Closing is expected in Q4 2026, which gives Naehas customers a short window to lock pricing, roadmap and data terms.

Protectt.ai, a Series A mobile app security startup, is taking control of Bosch's AIShield. Customers of AISpectra and AIGuardian should confirm their contracting entity, support team and on-prem roadmap now.

GXO will deploy Agility's Digit humanoids at Statesville, NC and Lebanon, TN in early 2027 on a robots-as-a-service contract, months after its CEO said 45 humanoid pilots have not reached ROI. The contract structure is the part worth copying.

Anthropic's S-1 shows $16.3B of H1 2026 revenue, a $974M Q2 operating profit and 48% of sales through cloud marketplaces. For Claude buyers, it shows which contract terms to lock in: version pins, notice periods, capacity and a fallback model.

Itaú Unibanco agreed to pay R$2.925 billion (about $589 million) for NeoSpace, the tabular foundation-model startup behind its credit, CRM and fraud models. Other NeoSpace customers now buy from a vendor a rival bank owns.

LSEG agreed to buy Octus for $3.535 billion in cash, putting the covenant and private credit data its 750 customers reach through the platform, an MCP connector and AI workflows inside the owner of Workspace. The release says nothing about pricing or contract terms, so credit teams should lock in terms before the H1 2027 close.

AAIF took in Open Responses, Agent Skills, OpenSandbox and OpenEnv at AGNTCon + MCPCon. Two come with tests or validators and named adopters; two are still early.

GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna now draw down Google Cloud commits, but Google's savings-plan terms give third-party models no discount, and the private preview has no published price, regions or ZDR yet.

OpenAI is moving Orchestra Bio's drug-program agents into Rosalind and has not said whether the standalone product or its contracts survive. Customers should export their data and get continuity terms in writing.

Azure and Google now meter reserved model capacity at pay-as-you-go token weights, so a monthly term needs about 99% utilization to save money. Annual terms break even at 84% on Azure and 75% on Google; Bedrock publishes too little to compute.

Salesforce is paying $47.50 a share in cash for Braze and keeping its brand and CEO. Pricing, warehouse connectors and standalone sales are not addressed, so Braze customers should negotiate before the deal closes in Salesforce's fiscal 2028.

Muse Spark Open is the closest US replacement for Kimi K3, five index points behind and cheaper per task. Defense, nuclear and critical-infrastructure buyers should pick Apache 2.0 Reflection Beam instead.

Copilot Studio agents with their own Entra account share one memory across every user, and Microsoft says it can surface file and chat summaries to people without access, regardless of sensitivity label. Admins can only scope who builds them and what they join.

Liquid AI's d1-ultra matches Jev's $0.042 per million input tokens, but it re-bills every image for every question, and its Decision Index lead of about eight points comes from text-only tests and more abstention.

Neptune's hosted users lost every run they had not exported in March 2026. Run MLflow on a database and bucket you own, keep W&B only with your own bucket, and schedule a weekly export whatever you pick.

Replit made OpenAI the default behind Replit Agent. OpenAI was already a listed subprocessor, so the DPA owes no notice, and enterprise admins cannot disable OpenAI entirely.

No CLM vendor has published extraction accuracy measured against a rival CLM. Shortlist Sirion and Icertis for a legacy repository, keep Agiloft for bespoke fields, and decide on a 200-contract bake-off scored per field.

DeepSeek's new Search & Fetch API costs $0.75 per 1,000 calls, about a tenth of Tavily's basic search, with billing from October 31. Queries go to a service that stores data in China under PRC-governed terms.

Agilent is buying Proscia for $240 million plus up to $30 million in milestones, and it already sells its own slide scanner. Labs and pharma teams on Concentriq should get scanner support, third-party AI access and data export written into their contracts before close.

Kimi K3 costs $3 / $15 at four of the five hosts compared that offer it and gpt-oss-120b $0.15 / $0.60 at four of six, but Moonshot's vendor audit found schema accuracy from 100% down to 71.96%. Fireworks is the default for agents; Together is the one to avoid for tool-heavy work.

Disney's $20M stalking-horse bid for Character.AI licenses code and patents only. User data, chats and trained models stay in the bankruptcy estate for any bidder by November 20.

Meta's 816B Muse Spark Open is free for internal use, but its license bans operating critical infrastructure and any nuclear or military use, the same clauses Llama 3 carried.

GTCR and Elliott agreed to take CCC private at a $6.1B enterprise value, with closing expected in Q1 2027. Carriers and repair shops running Estimate-STP and CCC ONE have until then to get price caps, data-use limits and exit rights into their contracts.

Cohere bills rerank per search and Voyage per token, so candidate count and chunk length decide the bill. Voyage rerank-3 is cheaper at most shapes; self-host BGE for residency or high volume.

Prime Intellect bought Moondream for its Photon engine, and Moondream Cloud and Lens close November 13. Base weights self-host free, but local fine-tunes still need a Moondream API key.

Colorado's ECDIS rules make your chief risk officer attest to the fairness framework, and the testing method is still unadopted. Run disparity tests on your own book with a named estimator and get remediation terms from the model vendor in writing.

Annex 22 and the FDA AI credibility guidance are still drafts on October 13, 2026. Buy GxP AI against Annex 11 and GAMP 5, test it to the draft Annex 22 on your own line, and keep a gap register for the rest.

Muse Wonder costs the same $4/$20 per million tokens as Claude Opus 5.5 and scores 52 to its 58 on Artificial Analysis, but runs about a quarter cheaper per task and leads on Harvey's legal agent benchmark. Wonder Full is reachable only through Muse Code subscriptions that train on your code by default.

Sopris Capital has closed its purchase of Blackford Analysis from Bayer, reported October 9, and is merging it into Azra AI. The only public commitment covers existing customers, with nothing on pricing, BAAs or the 45-plus AI vendors on the platform.

Nvidia is reportedly in talks to buy, acquihire or take a bigger stake in Reflection AI, and Beam's promised Apache 2.0 weights have not been published yet. Mirror a released checkpoint before you standardise on Beam.

Amazon is in talks to buy Decart for about $7 billion, which would put the DOS layer sold as neutral across Nvidia, Trainium and TPU inside Trainium's maker. Customers should get named multi-chip support in writing before a deal signs.

OpenAI told investors its run rate was near $50B at the end of September, about $20B below the circulating figure. Anthropic books partner sales gross and OpenAI books them net, so compare the two on one basis.

The Deno team is joining Cloudflare. Deno Deploy shuts down six months after the October 9 announcement, the runtime gets one more year of security releases, and Cloudflare's self-hosted Workers plan has no date.

Behavox has closed its purchase of Apiax, the Zurich engine that turns financial regulation into machine-readable rules. Clients are promised the same product and support, and nothing yet on price, roadmap, hosting or exit.

Microsoft's Decision-1 lists at $0.042 per million input tokens with free output, the same price as TypeSafe's Jev. Its 'highest accuracy' claim rests on 36 unnamed benchmarks with no published scores or calibration data.

For annual 10 m scoring, AlphaEarth embeddings plus a few hundred of your own labels match a bespoke imagery pipeline. Pin the version, validate with a geographic buffer, and keep the per-scene vendor only for daily or sub-metre work.

Under NYC Local Law 144 the employer owns the AI hiring bias audit, and a vendor's audit counts only if your applicant data went into it. Hire an auditor who runs the numbers on your own file.

micro1 will spend $1 billion over 12 months buying company SOPs, CRM data and decision records for AI training. Its public terms name no buyer, exclusivity or liability, and much of the data it wants carries customer and vendor rights.