Blackbaud launches Lantern fundraising language model
Howard Lake | 4 November 2026 | News
Blackbaud has unveiled Lantern, which it calls the first domain-specific language model purpose-built for fundraising. It will not reach customers until 2027, and the company has not said where it will be offered first.
Announced on 29 September 2026 at bbcon 2026 in Columbus, Ohio, Lantern™️ was developed in a strategic collaboration with data and AI company Databricks. It will sit in the intelligence layer of Platform for Good, the unified operating system that Blackbaud launched at the same event. Blackbaud says more than $100 billion is raised, granted or managed through its platforms every year.
Chief executive Mike Gianoni said general-purpose frontier models “learn from the internet, while Lantern learns from philanthropy.”
Advertisement
When and where Lantern will be available
Lantern will begin powering select Raiser’s Edge NXT and Agents for Good capabilities in 2027, with broader availability planned across Platform for Good. Blackbaud has not published a launch date, pricing or a list of countries. On its third quarter earnings call, management did not say whether availability would come in the first or second half of 2027.
Blackbaud has also outlined Lantern plans to Australian customers at bbcon Sydney. UK charity users should ask their account manager about timing, licensing and data hosting.
What Lantern was trained on
Blackbaud says Lantern combines leading open-weight models, running on the Databricks Data + AI Platform, with its own 40+ years of sector experience. The base models have not been named.
Lantern is then post-trained on “synthetic scenarios” modelled on Blackbaud’s Social Impact Signal Graph, designed to reflect real-world patterns such as donor retention, upgrade paths and lifetime relationships. On the earnings call, Gianoni said Lantern does not use customer-specific data. Instead it uses synthetically created data that mimics customer data, alongside Blackbaud’s own sector data assets, according to a transcript of the call.
Is Lantern unique?
Blackbaud’s claim to have developed the first fundraising sector-specific language model seems to be accurate, although sector-specific AI is now common in vendor marketing.
Bonterra says its Que engine is “trained exclusively on nonprofit and social impact data” and cites 1.16 billion network transactions, but presents Que as an agentic assistant built into its products rather than a separately named model.
GoFundMe Pro’s tools use machine learning trained on its own fundraising dataset, and Virtuous has launched an AI research agent for donor prospecting.
Sector response
Tim Cerato of Boston University, one of the customers consulted during development, said each industry has its own language, rules and guardrails that AI needs to reflect.
MarTech Edge said Lantern’s value will depend on real-world performance and measurable outcomes after deployment.
Blackbaud research on AI use
A Blackbaud Institute survey, conducted in the USA in March 2026 with 1,389 social impact professionals and 1,034 donors, found that 85 per cent of professionals use AI, but only about 10 per cent of organisations see significant returns.
The same research found that 76 per cent of donors say it is important for organisations to disclose when and how they use AI, while only 26 per cent of professionals say their organisation does so. That gap matters whichever vendor’s tools a charity adopts.
- Bridgit wins People’s Choice at Blackbaud’s bbcon (19 October 2026)
- Bridgit joins Blackbaud’s global accelerator (2 February 2026)
- Highlights from Blackbaud’s bbcon 2020 (9 October 2020)
Related posts
About Howard Lake
Howard Lake is consultant editor of UK Fundraising. He founded the site in 1994 and successfully sold it in 2022. As director of Giving X Ltd he is exploring growing giving on a massive scale.
He is the founder of Fundraising Camp and co-founder of GoodJobs.