Building AI-powered sales software with product integrations

In the last 6 months, we’ve seen a massive wave of AI-powered software flood the enterprise SaaS market. Nowhere has this been more evident than in sales software, a $100+ billion category where startups and incumbent leaders alike have been racing to build the next generation of AI-powered sales tools.

With foundation models like OpenAI’s GPT-4 quickly becoming commoditized, AI quality in sales software has largely become a function of the data used to train, tune, and prompt the models. Better input data leads to better performance and better product experiences for customers.

The rise of AI-powered sales software has brought the product integrations which supply key customer data inputs to the forefront for many companies. For developers, having a reliable solution for integrating with their customers’ CRMs and sales engagement platforms has become more important than ever.

How sales software companies use AI

AI has already enhanced many existing sales workflows through a combination of improved automation and personalization.

For example, companies like Coldreach.ai, OpnBx.ai, and Regie.ai use AI to improve the sales prospecting workflow by increasing outbound campaign conversion. Instead of relying on static email templates with placeholder tokens, these products are able to leverage powerful AI models to generate more personalized messages and replies.

Fathom, Gong, and Fireflies use AI to summarize and extract insights from call transcripts. These products reduce or eliminate the manual work involved in typing up and sharing call notes, and drafting follow-up emails for customers.

Madkudu and Correlated use AI to score and qualified leads, so sales teams can focus their efforts on engaging the right prospects at the right times. They and other revenue intelligence products leverage AI to improve sales targeting and efficiency across the entire sales lifecycle.

In some cases, AI has unlocked massive efficiency gains that just were had to imagine just a year ago. For example, Tavus and Reachout.ai use AI to generate personalized videos for tens of thousands of prospects at a time, a task that previously would not have been possible at this scale without a massive team of humans.

Integration patterns in AI-powered sales tools

User-facing product integrations are core to all of the use cases described above and many others. At Supaglue, we consistently see three main integration patterns in AI-powered sales tools:

Syncing data

Syncing data is one of the most common ways sales tools integrate with their customer’s CRMs. Many companies ingest their customers’ Contact, Account, and Opportunity records into an application database or data warehouse. The data is periodically refreshed on a schedule to reflect newly updated records. Synced data is often joined against other customer data or enriched with 3rd party data to power AI/ML models for analytics, lead scoring, and other revenue intelligence use cases.

Action triggers

Another common integration pattern is action triggers. From a sales tool, a customer can create a new record (e.g. add a new lead) or update a field in an existing record (e.g. change an account’s owner). They can add users to an existing email sequence or create new sequences with personalized, AI-generated emails. Action triggers are also used for CRM enrichment use cases, where lead scores and custom metrics are pushed back into the customers’ system of record, where they are more easily accessible for sales teams.

Real-time events

Data freshness is critical for some sales automation use cases. For example, in certain products, when the stage of an opportunity changes, customers need to get notified immediately when the change happens. Companies who need real-time data typically consume these changes as event streams via a webhook integration. When they receive a webhook, their products will trigger the appropriate internal workflow or notification in their products.

Integration considerations for developers

AI-powered sales tools need to answer a few fundamental questions about their product integration strategy:

Depth vs breadth

Depending on your product and what your customers are asking for, it may make more sense to build a large catalog of product integrations, or to build deeper integrations for a smaller number of providers. Do you need to support a long tail of CRMs, or is it more important to access more features within Salesforce? In our experience, given the customizability of CRMs, AI-powered sales tools tend to benefit from access to unstructured data, custom objects, customer-specific field mappings, and other advanced integration features.

Access patterns

We mentioned the 3 access patterns we see most often earlier. In practice, many companies start with one access pattern and expand to the others over time. One common path is for products to sync data more frequently as their customers become more sophisticated (e.g. every 1 hour → every 10 minutes), eventually converging to real-time events. Another common product evolution path is syncing data to provide a “read-only” intelligence application, and layering on action triggers to perform writes later. Which access patterns make the most sense for your application?

Maintenance costs

Finally, companies (especially startups) consistently underestimate the complexity of maintaining integrations with 3rd party systems that vary significantly across functionality, interfaces, and reliability. Add to the fact that each system has its own roadmap and evolves its API over time, and the cost of maintaining integrations adds up pretty quickly. If you’re looking to build integrations yourself, be conservative when budgeting development resources and expect there to be an ongoing maintenance tax. How much time and resources are you willing to invest in product integrations?

Accelerating integrations for AI-powered sales products

If AI will transform sales software, then product integrations will determine which AI-powered sales products win. However, there's a lot for developers to consider before building product integrations, and challenges will surface in the form of maintenance costs over time.

At Supaglue, we spend our days (and nights!) thinking about these problems. We're building an open source platform that helps developers ship the critical product integrations their AI-powered sales products need. If you’re interested in learning about how Supaglue can help accelerate your integrations roadmap, reach out to us at hello@supaglue.com.