Dhiren Bhatia, Vishakh Hegde and Gaurav Nemade: Building AI Agents to Transform Enterprise RFP Workflows

For many enterprise sales teams, winning a major customer begins with a long and demanding document.

A request for proposal, or RFP, may contain hundreds of questions about a company’s products, technical capabilities, pricing, security, legal policies, implementation process, and customer support. Completing the response can require input from sales, engineering, finance, legal, compliance, and product teams.

The information often exists somewhere inside the organization. The main challenge is not simply answering the question. Teams must find reliable information, confirm that it is current, tailor the response to the customer, and obtain approval before the deadline.

Dhiren Bhatia, Vishakh Hegde, and Gaurav Nemade founded Inventive AI to improve this process. Their San Francisco-based company develops AI agents that help businesses prepare RFPs, security questionnaires, requests for information, and other sales documents.

Inventive AI does more than generate text. Its platform connects with a company’s existing knowledge sources, prepares draft answers, identifies conflicting information, and supports collaboration across the response process.

For CEOs and sales leaders, the company represents a practical form of enterprise AI: technology designed around a specific workflow, measurable business pressure, and the knowledge that employees already use.

Why RFP Responses Are Difficult to Manage

RFPs can influence some of the largest deals in an enterprise sales pipeline. They also create a considerable amount of work.

A proposal manager may begin by dividing the document into sections. Security questions go to the information security team. Contract terms may need legal review. Product capabilities require input from specialists. Pricing moves to finance or sales leadership.

Answers from previous proposals can help, but they may be stored across shared drives, spreadsheets, email threads, content libraries, and older documents. Some statements may no longer reflect the current product.

As a result, employees spend valuable time searching for information and asking subject-matter experts to repeat answers they have provided before. The final response may then pass through several review cycles with inconsistent language or missing details.

This creates more than an administrative problem. A late or inaccurate response can damage the company’s credibility with a potential customer. In highly competitive sales processes, slow execution may also reduce the time available for strategy and customization.

Could generative AI write all the answers? It can produce fluent text, but fluency does not always mean correctness. Enterprise proposals need information grounded in approved company sources.

Inventive AI was built around that distinction.

Three Founders With Complementary Experience

The backgrounds of Bhatia, Hegde, and Nemade bring together entrepreneurship, enterprise software, machine learning, and AI product development.

Bhatia is Inventive AI’s co-founder and CEO. He previously founded Viewics, a healthcare analytics software company that Roche acquired in 2017. That experience gave him a close view of enterprise sales and the effort required to complete RFPs.

According to Sierra Ventures, Bhatia experienced the problem again while working as an executive at a Fortune 500 company. He saw sales and technical employees spending significant time on proposals (instead of customer strategy and other valuable work).

Hegde serves as co-founder and chief technology officer. Y Combinator reports that he previously led machine-learning work at Ambient.ai, studied AI and machine learning at Stanford, and completed his undergraduate education at the Indian Institute of Technology Madras. He has also published research and written a book about deep learning.

Nemade brings experience from Google AI and Google Brain. He previously worked as a product manager and was the first product manager for Google’s LaMDA platform, the language-model technology associated with the early development of Bard. He studied computer science at the Indian Institute of Technology Roorkee.

Together, the founders combined a clear business problem with the technical knowledge required to build a specialized AI product.

From Enterprise Frustration to Inventive AI

Inventive AI was founded in 2023 and joined Y Combinator’s Summer 2023 group.

The company focused on proposal workflows because they contain many tasks suited to AI assistance. Employees must search large collections of documents, compare similar answers, summarize technical details, draft new responses, and identify missing information.

However, the founders also recognized that an enterprise cannot allow a general-purpose AI tool to invent facts about security controls, product capabilities, or contractual commitments.

Their solution was to ground the system in the customer’s own information.

In August 2024, Inventive AI announced that it had raised $4 million in funding. Sierra Ventures led the investment, with participation from Y Combinator and other investors. The company said the funding would support product development and growth.

The investment reflected wider interest in AI products that address defined business workflows rather than offering a general chat interface.

Creating a Central Source for RFP Knowledge

A strong RFP response depends on reliable information. Inventive AI helps businesses collect proposal content from multiple sources in one place.

The platform can connect with services such as Google Drive, Microsoft SharePoint, Notion, Confluence, Salesforce, and Slack. Organizations can also upload previous RFPs, product documents, security policies, and approved question-and-answer libraries.

Once connected, the system can search these sources when preparing a response. The goal is to produce a draft based on company material rather than unsupported information from the wider internet.

This approach can reduce the time employees spend searching through folders or contacting multiple departments.

A central knowledge hub can also improve consistency. If several sales employees answer the same security question, they should not provide different descriptions of the company’s practices. Approved information gives the team a stronger starting point.

The system must still be managed carefully. A central library is useful only when it contains accurate information, has proper access controls, and allows authorised employees to find what they need.

How Inventive AI’s Agents Support the Workflow

Inventive AI uses several AI functions across the proposal process.

The platform first analyzes an RFP or questionnaire and helps organize the questions. It can then search the connected knowledge sources and prepare suggested answers.

Employees review those drafts, make changes, add customer-specific details, and involve specialists where necessary. This can reduce the amount of writing required while keeping people responsible for the final submission.

Inventive AI also provides tools that can:

• – Draft responses using approved company content
• – Find relevant information across connected systems
• – Flag questions that lack enough source material
• – Identify conflicting or outdated answers
• – Suggest improvements to clarity and positioning
• – Support collaboration and internal review
• – Research competitors for proposal strategy
• – Track content that may need an update

The company says its AI Content Manager scans knowledge sources for information that may be old or inconsistent. This is important because an outdated statement can easily appear in multiple proposals when employees continue copying it from earlier documents.

Instead of treating content management as a separate task, Inventive AI connects it with daily proposal work.

Moving From Automation to AI Agents

Traditional RFP software often relies on a library of saved questions and answers. When a new request arrives, the platform searches for a similar question and recommends a stored response.

This can save time, but the method depends on employees maintaining the library. It may also struggle when a buyer asks a familiar question in a new way or expects an answer tailored to its industry.

Inventive AI’s agent-based approach is designed to handle more of the surrounding work. An AI agent can gather information from several sources, understand the question, prepare a draft, identify missing evidence, and improve the response before it is sent.

The difference is not complete independence. The system is still most useful when employees review its work and approve important statements.

AI agents can complete several connected steps, but leaders must decide where automation ends and human responsibility begins.

For a proposal involving legal commitments, security practices, pricing, or product roadmaps, approval from the appropriate department remains essential.

Giving Experts More Time for Strategic Work

Subject-matter experts are important to the RFP process, but answering questionnaires is rarely their main job.

A security leader may receive the same questions about encryption, access controls, or incident response in many different formats. A product manager may repeatedly explain a feature that is already documented. Each interruption takes time away from other responsibilities.

AI can prepare an initial answer using approved information. The expert can then focus on checking accuracy, handling unusual questions, and explaining areas that require judgment.

This changes the role of specialists from writing every response to reviewing the most important content.

For sales teams, faster access to knowledge creates more time to improve the proposal’s strategy. They can focus on the customer’s goals, the competitive situation, and the value their solution provides.

This creates business value beyond simply reducing administrative work. A better process may allow teams to pursue more opportunities without reducing the attention given to each buyer.

Measuring Results Without Ignoring Context

Inventive AI states that its platform can make RFP response workflows more than 70% more efficient. It also promotes faster draft creation and shorter completion times.

The company features customer accounts from organizations including AssetWorks, HiBob, Rad AI, and others. According to an AssetWorks representative, the company reduced its response time by 90%. Other customers describe handling more proposals or reducing the time spent searching for content.

These figures come from Inventive AI and its customers. They show what individual organizations have experienced, but they should not be treated as guaranteed results.

Performance will depend on several factors:

• – The quality of the existing content
• – The number and complexity of RFP questions
• – Integration with company systems
• – Employee adoption
• – Review and approval requirements
• – The amount of customization expected by each buyer

Executives evaluating an AI platform should begin with a defined pilot. They can compare response time, review cycles, employee hours, answer quality, and proposal volume before expanding the system.

Accuracy, Security, and Human Review

RFPs often contain confidential business information. A questionnaire may ask for technical architecture, cybersecurity practices, customer references, financial details, or internal policies.

Companies must understand how an AI provider handles this data. Their review should cover encryption, user permissions, retention policies, third-party access, audit controls, and the use of customer information for model training.

Inventive AI states that it uses role-based access controls and encryption. It also reports SOC 2 compliance and support for privacy requirements.

The platform shows how confident it is in each answer and alerts users when there is not enough information to provide a reliable response. This is safer than generating a response without evidence, but human review remains necessary.

A polished answer can still be incorrect if the source document is old or if the question requires legal interpretation. Businesses should assign clear owners for content approval and create rules for sensitive sections.

What Leaders Can Learn From the Founding Team

Bhatia, Hegde, and Nemade offer several useful lessons for enterprise leaders.

First, successful AI products often begin with a narrow and expensive problem. RFP work involves clear deadlines, repeated tasks, and employees from several departments. That makes the impact easier to measure.

Second, domain experience matters. Bhatia understood the problem as a founder and enterprise executive. Hegde and Nemade brought the technical background needed to apply AI without ignoring reliability.

Third, company knowledge is a competitive asset. AI becomes more useful when it can work with accurate internal information instead of producing generic content.

Finally, automation should support expert judgment. The platform creates speed by preparing drafts and finding answers, while employees remain responsible for strategy, approval, and customer commitments.

Transforming RFPs Into a Strategic Process

RFPs will remain demanding because enterprise buyers need detailed evidence before making major purchases. AI cannot remove the need for security reviews, legal approval, technical expertise, or careful customer communication.

What it can change is the amount of time spent repeating earlier work.

Inventive AI brings the proposal process into one workflow, allowing teams to find existing information, draft initial responses, identify content gaps, and manage reviews. This allows employees to spend less time searching through documents and more time strengthening the proposal.

Dhiren Bhatia, Vishakh Hegde, and Gaurav Nemade are building their company around a practical idea: enterprise AI should help people complete important work with greater speed and control.

For CEOs, founders, and managing directors, that approach offers a broader lesson. The greatest value from AI may not come from replacing an entire department. It may come from improving a critical process that has remained slow for years.