Cimba.AI emerges with $1.25M to help enterprises build AI agents

OpenAI made global headlines last year with its launch of an easy, natural language tool for building customized chatbots called GPTs, an approach imitated months later by the open source AI platform Hugging Face with its Hugging Chat Assistants.

And while both of those tools are already being used by early adopters, it remains to be seen if enterprise customers hop aboard, and if so, how many.

But said enterprise customers may want to consider yet another option, a new generative AI startup dedicated specifically to this task: Cimba.AI, founded by former Airbnb and AWS veterans, which emerged from stealth today with $1.25 million in pre-seed funding led by Ripple Ventures, joined by SeaChange, PackVC, and angel investors including Chad Sanderson and Chris Riccomini. 

Headquartered in Seattle, Cimba’s web-based platform allows enterprises to create custom AI agents to aid employees in their workflows; pull data in useful forms like dashboards and graphs; and streamline recurring business operations, drawing upon each enterprise’s distinct, relevant, contextual and proprietary data.

“Our goal is to help businesses build their own Jarvis for their bigger operation, which can give you an insight as well as taking action,” said Subrata (Subu) Biswas, Cimba’s co-founder & CEO, in a video call interview with VentureBeat.

The name Cimba is actually an acronym that stands for “Contextualized Insights Make Business Action,” Biswas wrote in a LinkedIn post today.

No price was publicly listed by the company for its inaugural offering, but Biswas said it was “typical software-as-a-service (SaaS) subscription” based pricing per seat, depending on the number of users the enterprise customer sought.

AI-driven business operations

Co-founded by Biswas, a former engineering manager at Airbnb, and Vishal Das, a former applied scientist at AWS, Cimba.AI allows enterprise employees without any technical skill to build self-learning AI agents tailored to their organization’s unique knowledge base in minutes rather than the hours or days and skilled developer labor needed prior to the Gen AI era.

“We are taking out 95% of the work that a data scientist would have to do to create this agent by themselves,” Biswas told VentureBeat.

The Cimba AI Agent builder tool looks similar to the GPT Builder from OpenAI and Hugging Chat Assistants builder from Hugging Face, complete with several text fields where an enterprise employee can describe in plain, conversational English the kind of functionality they want their AI agent to perform.

Then, they can also specify where the AI Agent should surface or draw information from. This includes integrating dashboards, metadata, playbooks and query history from databases like Snowflake and Amazon Redshift, as well as files from Salesforce, Hubspot, Google Drive and more. At present, the Cimba.ai platform supports dozens of integrations with common enterprise APIs.

Once they’ve done all that, Cimba’s web interface simply puts together the agent into a neat and tidy button, and the user can try it out and refine it as needed.

Through these AI Agents, Cimba claims to offer the ability to understand complex business goals posed in natural language and recommend optimized workflows.

Empowering data scientists and those who demand on them on other enterprise teams

Biswas told VentureBeat that the tool was designed for existing data scientists to be able to easily create AI agents on demand for other departments, from sales to marketing, but that anyone who wanted could theoretically try it out and use it — though they would likely have a harder time refining and troubleshooting it.

AI agents can be created for separate departments or groups, job functions, or even individual users, and sharing permissions and access updated accordingly from Cimba’s web interface.

For example, Biswas mentioned that a marketing department could easily rely on a custom Cimba AI agent to answer the question “Which customers should I focus on this week?”

The Cimba-created agent would not only answer with the result based on the latest data of that enterprise customer, but also provide suggestions and next steps for the best type of outreach to make based on data.

“Once you hit ‘confirm’ it will recommend next actions and can even take some of them for you,” Biswas said, such as drafting personalized marketing emails or messages to customers.

Built atop existing foundation models with added safeguards

While Cimba has created an impressive engine for converting natural language into working, enterprise-grade AI assistants, it relies on others in the wider Gen AI marketplace to help power its capabilities.

Biswas told VentureBeat that Cimba is specifically using OpenAI’s GPT models and Anthropic’s Claude models at “different stages” depending on what the AI agent is being asked to do.

As for how Cimba handles sensitive and proprietary customer data surfaced and used by the resulting AI agents, Biswas said that the company creates a cache and also stores it in a vector database it has built using Pinecone, which is partitioned for every enterprise customer.

He also noted that Cimba’s “systems are flexible” and can support data being stored in a virtual private cloud for specific enterprises that seek this additional measure.

Key endorsements

Though it is early in Cimba’s story as a company, this enterprise Gen AI startup has already won the support of some big names and of course, their funding.

“Their approach to AI-driven business operations is not just innovative; it’s the future,” said Ripple Ventures Partner Dom Lau.

“Cimba.AI’s strength in automating repetitive data analysis tasks, thereby boosting productivity and efficiency,” noted Vineeth Loganathan, director of data science at ViralGains.

Several design partners also chimed in on Biswas’s LinkedIn post announcing the pre-seed funding: “Congratulations…I am so excited for you guys! Your product vision is truly innovative and ground breaking! Keep up the good job!” wrote Lior Solomon, VP of Data at Drata.

“Thrilled to see the product launch! As we’ve discussed, its potential is enormous. Wishing you all the best and keeping my fingers crossed for your success,” added Michał Szczudlak, Data Team Manager at Profitroom.

Currently in private beta, Cimba.AI is already working with mid-large enterprises. Its self-service platform enables non-technical users to easily build AI agents tailored to their business needs. With its innovative approach, Cimba.AI promises to reshape how organizations leverage AI to transform operations and decision making